An evaluation of selected (Q)SARs/expert systems for predicting skin sensitisation potential

Predictive testing to characterise substances for their skin sensitisation potential has historically been based on animal models such as the Local Lymph Node Assay (LLNA) and the Guinea Pig Maximisation Test (GPMT). In recent years, EU regulations, have provided a strong incentive to develop non-animal alternatives, such as expert systems software. Here we selected three different types of expert systems: VEGA (statistical), Derek Nexus (knowledge-based) and TIMES-SS (hybrid), and evaluated their performance using two large sets of animal data: one set of 1249 substances from eChemportal and a second set of 515 substances from NICEATM. A model was considered successful at predicting skin sensitisation potential if it had at least the same balanced accuracy as the LLNA and the GPMT had in predicting the other outcomes, which ranged from 79% to 86%. We found that the highest balanced accuracy of any of the expert systems evaluated was 65% when making global predictions. For substances within the domain of TIMES-SS, however, balanced accuracies for the two datasets were found to be 79% and 82%. In those cases where a chemical was within the TIMES-SS domain, the TIMES-SS skin sensitisation hazard prediction had the same confidence as the result from LLNA or GPMT.

An Automated and Intelligent Medical Decision Support System for Brain MRI Scans Classification

A wide interest has been observed in the medical health care applications that interpret neuroimaging scans by machine learning systems. This research proposes an intelligent, automatic, accurate, and robust classification technique to classify the human brain magnetic resonance image (MRI) as normal or abnormal, to cater down the human error during identifying the diseases in brain MRIs. In this study, fast discrete wavelet transform (DWT), principal component analysis (PCA), and least squares support vector machine (LS-SVM) are used as basic components. Firstly, fast DWT is employed to extract the salient features of brain MRI, followed by PCA, which reduces the dimensions of the features. These reduced feature vectors also shrink the memory storage consumption by 99.5%. At last, an advanced classification technique based on LS-SVM is applied to brain MR image classification using reduced features. For improving the efficiency, LS-SVM is used with non-linear radial basis function (RBF) kernel. The proposed algorithm intelligently determines the optimized values of the hyper-parameters of the RBF kernel and also applied k-fold stratified cross validation to enhance the generalization of the system. The method was tested by 340 patients’ benchmark datasets of T1-weighted and T2-weighted scans. From the analysis of experimental results and performance comparisons, it is observed that the proposed medical decision support system outperformed all other modern classifiers and achieves 100% accuracy rate (specificity/sensitivity 100%/100%). Furthermore, in terms of computation time, the proposed technique is significantly faster than the recent well-known methods, and it improves the efficiency by 71%, 3%, and 4% on feature extraction stage, feature reduction stage, and classification stage, respectively. These results indicate that the proposed well-trained machine learning system has the potential to make accurate predictions about brain abnormalities from the individual subjects, therefore, it can be used as a significant tool in clinical practice.

Open Ambient Intelligence Environments

The present impact of ambient intelligence concepts in eInclusion is first briefly reviewed. Suggestions and examples of how ambient intelligent environments should be specified, designed and used to favour independent living of people with activity limitations are presented.

[Automation expert system for quantitative risk assessment in reconstructive surgical procedures on the large intestine]

The algorithm of quantitative estimation of operative risk while performing reconstructive-restoration colonic surgery was proposed. There was established high sensitivity and specificity of quantitative estimation of operative risk while application of a neuronet models, especially in usage of a neuronet owing two output neurons. While doing the operative risk estimation, using neuronets, the fault answers rate, as well as of unrecognized states, is trustworthy small, what permits to recommend the method for wide application.

The use of point-of-sale machines in school cafeterias as a method of parental influence over child lunch food choices

BACKGROUND

Computerized point-of-sale (POS) machine software that allows parents to place restrictions on their child's school meal accounts is available. Parents could restrict specific foods (eg, chips), identify specific days the child can purchase extra foods, or set monetary limits. This descriptive study examines the use of parental restrictions on student cafeteria POS accounts in a convenience sample of 2 school districts.

METHODS

POS alerts, with student gender, grade, ethnicity, and students' free or reduced-price meal eligibility, were obtained from 2 school food service departments for the 2007-2008 school year. The alerts were coded into 5 categories: financial, medical, restrictions, snacks OK, and extras OK. The distribution of alerts by district, students, and demographics was then tabulated.

RESULTS

District A (4839 students) had more students with alerts (n = 789, 16%) than District B (8510 students; n = 217, 2.6%), and 94 District A students had a second alert. District A parents had to provide written permission for their child to purchase snacks (n = 654, 13.5%) and extra meal items (n = 113, 2.3%). Most alerts were for full-pay students in both districts (74% and 66%) and varied by demographics of the students.

CONCLUSIONS

Few parents actually used this system to limit student purchases of foods outside the school meal. Future studies should investigate the influence of these restrictions on student food choices.

Use of an expert concept map as an advance organizer to improve understanding of respiratory failure

BACKGROUND

Helping novices transition toward expertise requires "meaningful" learning. Advance organizers are educational tools which help connect prior knowledge with new information, a critical step in making learning meaningful. Concept maps visually represent knowledge organization and can serve as advance organizers enabling deeper and more meaningful learning while enhancing knowledge integration.

AIM

To compare respiratory failure understanding of resident physicians instructed, using an expert concept map advance organizer with learners receiving traditional didactic teaching.

METHODS

Residents were randomized by month of service to receive either a control lecture or a session using an expert concept map as an advanced organizer. Participants completed three concept maps; pre-education (CM1), immediately post-education (CM2), and 1 week later (CM3). Concept maps were scored using a standardized structural scoring method.

RESULTS

Forty-six pediatric residents (23 control and 23 experimental) participated. To account for repeated measures within subjects, the generalized estimating equations method compared concept map improvement between groups. The experimental group improved significantly more than controls (CM1-CM2-CM3 p = 0.001; CM1-CM2 p = 0.001; and CM1-CM3 p = 0.017).

CONCLUSIONS

Using an expert concept map as an advance organizer improves knowledge organization and integration while offering a tool to enhance deeper understanding of medical knowledge among resident physicians.

Medical decision-support systems and the concept of context

Medical decision-support systems are of necessity multi-contextual in nature. There are always at least two contexts involved in the use of such systems: the expert knowledge-provider context and the end-user context. To show this, we present examples of context-dependent aspects significant to the use of decision-support systems. The existence of discrepancies between the contexts threatens to disrupt the rationale for using decision-support systems: for the system to transfer knowledge from the expert to the end-user. Both theoretical and empirical studies show that such discrepancies exist and that they may be detrimental to the use of decision-support systems. Systems must thus give support in interpreting the output produced by the system in the context of the end-user.

Understanding ethics guidelines using an internet-based expert system

National and international guidelines outlining ethical conduct in research involving humans and animals have evolved into large and complex documents making the process of gaining ethics approval a complicated task for researchers in the area. Researchers, in particular those who are relatively new to the ethics approval process, can struggle to understand the parts of an ethics guideline that apply to their research and the nature of their ethical obligations to trial participants. With the scope of medical research likely to continue to expand in the future, it is clear that ethics guidelines will only increase in complexity and number. This paper describes one possible solution to the

PROBLEM

the use of an internet-based expert system to intelligently and interactively distribute the information stored in ethics guidelines to individual researchers. This paper also details how one such system was designed and tested with respect to Australian medical research ethics guidelines.

[New technologies in the optimization of CRT programming]

After implanting a CRT device, consistent and scheduled patient follow-up is mandatory. Besides determining electrode parameters and reviewing arrhythmic episodes, these follow-ups focus on monitoring and optimizing congestive heart failure therapy. Therefore new CRT devices present methods for heart failure surveillance and telemetric transmission of the acquired data, which allows the physician to respond immediately to the varying needs of the respective heart failure patient. In addition to cardiac resynchronization, optimization of atrioventricular (AV) and interventricular (VV) delay provide major hemodynamic benefits. As echocardiographic optimization of AV and VV delay is time consuming it is often not feasible during daily clinical practice. Therefore implemented algorithms that automatically determine and adapt AV and VV delays with respect to the fluctuating needs of the patients are essential. This article presents the current state of monitoring and optimization methods in CRT devices.

Acceptability of a stage-matched expert system intervention for domestic violence offenders

Most interventions for men who batter are standardized and "one-size-fits-all," neglecting individual differences in readiness to change. A multimedia expert system intervention based on the transtheoretical model (the "stage model") was developed as an adjunct to traditional court-mandated programs. The expert system assesses stage of change, decisional balance, self-efficacy, and processes of change and provides immediate individualized stage-matched feedback designed to increase readiness to end the violence. Fifty-eight male batterer intervention program clients were invited by agency staff to complete an expert system session and an evaluation of the program; 33 men were recruited at program intake and the remainder from ongoing groups. Responses to the intervention were very positive. For example, 87% of participants reported that they found the program to be easy to use, and 98% said it could probably or definitely help them change their attitudes or behaviors. Findings provide encouraging evidence of the acceptability of this stage-matched approach to intervention for domestic violence offenders.

Direct-to-patient expert system and home INR monitoring improves control of oral anticoagulation

BACKGROUND AND OBJECTIVE

Internet-based disease management programs have the potential to improve patient care. The objective of this study was to determine whether an interactive, internet-based system enabling supervised, patient self-management of oral anticoagulant therapy provided management comparable to an established anticoagulation clinic.

PATIENTS/METHODS

Sixty patients receiving chronic oral anticoagulant therapy who had access to the internet and a printer, were enrolled into this prospective, single-group, before-after study from a single clinic and managed between March 2002 and January 2003. Patients learned how to use a home prothrombin time monitor and how to access the system through the internet. Patients used the system for six months, with daily review by the supervising physician. The primary outcome variable was the difference in time in therapeutic range prior to and following introduction of internet-supervised patient self-management.

RESULTS

The mean time in therapeutic range increased from 63% in the anticoagulation clinic (control period) to 74.4% during internet-supervised patient self-management (study period). The mean difference score between control and study periods was 11.4% (P = 0.004, 95% confidence interval 5.5-17.3%). There were no hemorrhagic or thromboembolic complications.

CONCLUSIONS

This novel approach of internet-supervised patient self-management improved time in therapeutic range compared to an anticoagulation clinic. This is the first demonstration of an internet-based expert system enabling remote and effective management of patients on oral anticoagulants. Expert systems may be applicable for management of other chronic diseases.

[Remote control for magnetic resonance imaging as a part of daily routine examinations]

Continuous improvements in MR technology have lead to an increased number of clinical indications, such as whole-body MRI, as well as demanding examination protocols. A remote control (syngo Expert-i, Siemens Medical Solutions) allows radiologists and other experts to support the scan at any time without being physically present at the MR site. Consequently a high level of expertise is available for every exam. The radiologist can simply log on from any networked PC and a full-screen display with total mouse control within the MR console's user interface is provided. A local user can be advised with respect to the imaging strategy while the patient is still inside the scanner, thus resulting in a more time-efficient high-quality examination protocol. This paper provides an account of our initial experiences using the remote control and describes practical future applications.

A software engine to justify the conclusions of an expert system for detecting renal obstruction on 99mTc-MAG3 scans

The purposes of this study were to describe and evaluate a software engine to justify the conclusions reached by a renal expert system (RENEX) for assessing patients with suspected renal obstruction and to obtain from this evaluation new knowledge that can be incorporated into RENEX to attempt to improve diagnostic performance.

METHODS

RENEX consists of 60 heuristic rules extracted from the rules used by a domain expert to generate the knowledge base and a forward-chaining inference engine to determine obstruction. The justification engine keeps track of the sequence of the rules that are instantiated to reach a conclusion. The interpreter can then request justification by clicking on the specific conclusion. The justification process then reports the English translation of all concatenated rules instantiated to reach that conclusion. The justification engine was evaluated with a prospective group of 60 patients (117 kidneys). After reviewing the standard renal mercaptoacetyltriglycine (MAG3) scans obtained before and after the administration of furosemide, a masked expert determined whether each kidney was obstructed, whether the results were equivocal, or whether the kidney was not obstructed and identified and ranked the main variables associated with each interpretation. Two parameters were then tabulated: the frequency with which the main variables associated with obstruction by the expert were also justified by RENEX and the frequency with which the justification rules provided by RENEX were deemed to be correct by the expert. Only when RENEX and the domain expert agreed on the diagnosis (87 kidneys) were the results used to test the justification.

RESULTS

RENEX agreed with 91% (184/203) of the rules supplied by the expert for justifying the diagnosis. RENEX provided 103 additional rules justifying the diagnosis; the expert agreed that 102 (99%) were correct, although the rules were considered to be of secondary importance.

CONCLUSION

We have described and evaluated a software engine to justify the conclusions of RENEX for detecting renal obstruction with MAG3 renal scans obtained before and after the administration of furosemide. This tool is expected to increase physician confidence in the interpretations provided by RENEX and to assist physicians and trainees in gaining a higher level of expertise.

The role of telemedicine in CPAP compliance for patients with obstructive sleep apnea syndrome

The objective of this study was to compare continuous positive airway pressure (CPAP) use, functional status, and client satisfaction in obstructive sleep apnea syndrome (OSAS) patients randomized to either telemedicine support or traditional care. In our university-affiliated sleep disorders center, patients with OSAS who were initiating CPAP therapy were randomized to receive telemedicine support vs traditional follow-up care for 30 days. The telemedicine group received a "Health Buddy" computer that provided daily Internet-based informational support and feedback for problems experienced with CPAP use. At 30 days, there were no significant differences in the hours of CPAP use between groups receiving traditional care (M = 4.22, SD+/-2.05) and telemedicine support (M = 4.29, SD+/-2.15), p = 0.87, or in the proportion of nights with CPAP use between the traditional (M = 50%+/-33.8) and telemedicine groups (M = 47%+/-34.2), p=0.61. No significant differences were found between groups in functional status (M = 2.27+/-4.56 vs M = 2.03+/-3.88, respectively, p = 0.76) or client satisfaction (M = 28.0+/-3.51 vs M = 28.5+/-3.05, p = 0.43). Patients in the telemedicine and traditional groups had similar CPAP use, functional status, and client satisfaction. The data suggest that telemedicine support as provided by our model compares favorably with traditional care. As a provider-extender, telemedicine support for patients initiating use of CPAP may allow for greater practice efficiency while maintaining quality of care.

Advanced modes of mechanical ventilation: implications for practice

Mechanical ventilation is one of the most commonly applied interventions in intensive care units. Despite its life-saving role, mechanical ventilation is associated with additional risks to the patient and additional healthcare costs if not applied appropriately. To decrease risk, new ventilator modes continue to be developed with the goal of improving patient outcomes. Advances in ventilator modes include dual control modes that enable guaranteed tidal volume and inspiratory pressure, pressure-style modes that permit spontaneous breathing at high- and low-pressure levels, and closed-loop systems that facilitate ventilator manipulation of variables based on measured respiratory parameters. Clinicians need to develop a thorough understanding of these modes including their effects on underlying respiratory physiology to be able to deliver safe and appropriate patient care.

Automatic generation of spoken dialogue from medical plans and ontologies

This paper presents some research undertaken as part of the EU-funded HOMEY project, into the application of intelligent dialogue systems to healthcare systems. The work presented here concentrates on the ways in which knowledge of underlying task structure (e.g., a medical guideline) can be combined with ontological knowledge (e.g., medical semantic dictionaries) to provide a basis for the automatic generation of flexible and re-configurable dialogue. This approach is next evaluated via a specific application that provides decision support to general practitioners to help determine whether or not a patient should be referred to a cancer specialist. The competence of the resulting dialogue application, its speech recognition performance, and dialogue performance are all evaluated to determine the applicability of this approach.

An expert system for differential diagnosis of tall stature syndrome

There was created medical expert system for differential diagnosis of disorders and diseases manifested by tall stature. They were selected based on the information provided by two major computer databases, LDDB [London Dysmorphology Database] and Orphanet. Clinical signs, i.e. diagnostic criteria were developed according to 7 experts: manuals and textbooks, computer databases and online resources in pediatrics and rare diseases. Linguistic terms expressing the frequency and/or probability of presence of various symptoms were matched to numerical equivalents. The data from different experts were summarized according to the expertons' method and the inference engine was based on the Bayes theorem. An interface was made up by a set of slides with questions accompanied by boxes beside and user is expected to check corresponding boxes. The program was created in Borland C++ Builder. After having processed the entered data, the expert system produces the most probable five diagnostic possibilities and ranks them in order of likelihood.

Context-aware workflow management of mobile health applications

We propose a medical application management architecture that allows medical (IT) experts readily designing, developing and deploying context-aware mobile health (m-health) applications or services. In particular, we elaborate on how our application workflow management architecture enables chaining, coordinating, composing, and adapting context-sensitive medical application components such that critical Quality of Service (QoS) and Quality of Context (QoC) requirements typical for m-health applications or services can be met. This functional architectural support requires learning modules for distilling application-critical selection of attention and anticipation models. These models will help medical experts constructing and adjusting on-the-fly m-health application workflows and workflow strategies. We illustrate our context-aware workflow management paradigm for a m-health data delivery problem, in which optimal communication network configurations have to be determined.

Hypertext atlas of dermatopathology with expert system for epithelial tumors of the skin

BACKGROUND

Image resolution required for reference histological images is high. Obtaining high-resolution images requires a system, composing large image from individual smaller components. Such systems must thus be capable of automatically taking individual image parts, performing shading compensation and fusion of the image parts into one large image. Distribution of such images over the Internet requires developing a suitable user's interface with access to the image details.

METHODS

The ways of creating high-resolution images (virtual slides) and the interface enabling access to image details using Internet browser are described.

RESULTS

A collection of about 3200 dermatopathological, mostly histologic images is available at http://www.muni.cz/atlases: Hypertext atlas of dermatopathology. Methods of high-resolution image acquisition were used in digitizing the collection of skin lymphomas (Dermatology Institute, University Hospital, Zurich), which is a part of the atlas. The atlas is continuously maintained and upgraded; the number of contributors is growing.

Differentiating vascular pathophysiological states by objective analysis of flow dynamics

BACKGROUND AND PURPOSE

There is an unmet need to classify cerebrovascular conditions physiologically and to assess cerebrovascular system performance. The authors hypothesized that by simultaneously considering the dynamic parameters of flow velocity, acceleration, and pulsatility index (PI) (impedance) in individual Doppler spectrum waveforms, they could develop an objective method to elucidate the pathophysiology of vascular conditions and classify cerebrovascular disorders. This method, dynamic vascular analysis (DVA), is described.

METHODS

First, a theoretical model was developed to determine how any vascular segment and the ensemble of intracranial vascular segments could be defined according to its dynamic physiological characteristics. Next, the DVA method was applied to 847 anonymous serial complete clinical transcranial Doppler (TCD) studies of patients without regard for their diagnosis to ascertain actual reference ranges and the normality of the distribution curves for each dimension of the 3-parameter nomogram. The authors applied DVA to 2 clinical cases to see if they could track the changes in vascular performance of 2 known progressive diseases.

RESULTS

The theoretical analysis identified 295,245 possible vascular states for the ensemble of vascular segments in the cerebral circulation. When applied to clinical TCD data, DVA revealed continuous, normally distributed data for the velocity, PI, and logarithm of the acceleration.

CONCLUSIONS

DVA is proposed as a method for monitoring the physiological state of each cerebral artery segment individually and in ensemble. DVA evaluates the relationship among acceleration (force or pressure), velocity, and PI and provides an objective means to evaluate intracranial vascular segments using the paradigm of the well-described pressure-perfusion autoregulation relationship. DVA may be used to study cerebrovascular pathophysiology and to classify, evaluate, and monitor cerebrovascular disorders or systemic disorders with cerebrovascular effects.

Intelligent real-time performance monitoring and quality prediction for batch/fed-batch cultivations

Supervision of batch bioprocess operations in real-time during the progress of a batch run offers many advantages over end-of-batch quality control. Multivariate statistical techniques such as multiway partial least squares (MPLS) provide an efficient modeling and supervision framework. A new type of MPLS modeling technique that is especially suitable for online real-time process monitoring and the multivariate monitoring charts are presented. This online process monitoring technique is also extended to include predictions of end-of-batch quality measurements during the progress of a batch run. Process monitoring, quality estimation and fault diagnosis activities are automated and supervised by embedding them into a real-time knowledge-based system (RTKBS). Interpretation of multivariate charts is also automated through a generic rule-base for efficient alarm handling. The integrated RTKBS and the implementation of MPLS-based process monitoring and quality control are illustrated using a fed-batch penicillin production benchmark process simulator.

Learning to live independently with expert systems in memory rehabilitation

Expert systems (ES), which are a branch of artificial intelligence, has been widely used in different applications, including medical consultation and more recently in rehabilitation for assessment and intervention. The development and validation of an expert system for memory rehabilitation (ES-MR) is reported here. Through a web-based platform, ES-MR can provide experts with better decision making in providing intervention for persons with brain injuries, stroke, and dementia. The application and possible commercial production of a simultaneously developed version for "non-expert" users is proposed. This is especially useful for providing remote assistance to persons with permanent memory impairment when they reach a plateau of cognitive training and demand a prosthetic system to enhance memory for day-to-day independence. The potential use of ES-MR as a cognitive aid in conjunction with WAP mobile phones, Bluetooth technology, and Personal Digital Assistants (PDAs) is suggested as an avenue for future study.

Ambient intelligence in health care

Ambient Intelligence (AmI) is a new paradigm in information technology, in which people are empowered through a digital environment that is aware of their presence and context, and is sensitive, adaptive, and responsive to their needs, habits, gestures and emotions. The most ambitious expression of AmI is Intelligent Mixed Reality (IMR), an evolution of traditional virtual reality environments. Using IMR, it is possible to integrate computer interfaces into the real environment, so that the user can interact with other individuals and with the environment itself in the most natural and intuitive way. How does the emergence of the AmI paradigm influence the future of health care? Using a scenario-based approach, this paper outlines the possible role of AmI in health care by focusing on both its technological and relational nature. In this sense, clinicians and health care providers that want to exploit AmI potential need a significant attention to technology, ergonomics, project management, human factors and organizational changes in the structure of the relevant health service.

An expert system for fault management assistance on a space sleep experiment

The expert system, Principal Investigator-in-a-box, or [PI], was designed to assist astronauts or other operators in performing experiments outside their expertise. Currently, the software helps astronauts calibrate instruments for a Sleep and Respiration Experiment without contact with the investigator on the ground. It flew on the Space Shuttle missions STS-90 and STS-95. [PI] displays electrophysiological signals in real time, alerts astronauts via the indicator lights when a poor signal quality is detected, and advises astronauts how to restore good signal quality. Thirty subjects received training on the sleep instrumentation and the [PI] interface. A beneficial effects of [PI] and training reduced troubleshooting time. [PI] benefited subjects on the most difficult scenarios, even though its lights were not 100% accurate. Further, questionnaires showed that most subjects preferred monitoring waveforms with [PI] assistance rather than monitoring waveforms alone. This study addresses problems of complex troubleshooting and the extended time between training and execution that is common to many human operator situations on earth such as in power plant operation, and marine exploration.

Prevalence of restless legs syndrome and periodic limb movement disorder in the general population

BACKGROUND

Periodic limb movement disorder (PLMD) and restless legs syndrome (RLS) are two sleep disorders characterized by abnormal leg movements and are responsible for deterioration in sleep quality. However, the prevalence of these disorders is not well known in the general population. This study aims to document the prevalence of RLS and PLMD in the general population and to identify factors associated with these conditions.

METHODS

Cross-sectional studies were performed in the UK, Germany, Italy, Portugal and Spain. Overall, 18,980 subjects aged 15 to 100 years old representative of the general population of these five European countries underwent telephone interviews with the Sleep-EVAL system. A section of the questionnaire assessed leg symptoms during sleep. The diagnoses of PLMD and RLS were based on the minimal criteria provided by the International Classification of Sleep Disorders.

RESULTS

The prevalence of PLMD was 3.9% and RLS was 5.5%. RLS and PLMD were higher in women than in men. The prevalence of RLS significantly increased with age. In multivariate models, being a woman, the presence of musculoskeletal disease, heart disease, obstructive sleep apnea syndrome, cataplexy, doing physical activities close to bedtime and the presence of a mental disorder were significantly associated with both disorders. Factors specific to PLMD were: being a shift or night worker, snoring, daily coffee intake, use of hypnotics and stress. Factors solely associated with RLS were: advanced age, obesity, hypertension, loud snoring, drinking at least three alcoholic beverages per day, smoking more than 20 cigarettes per day and use of SSRI.

CONCLUSIONS

PLMD and RLS are prevalent in the general population. Both conditions are associated with several physical and mental disorders and may negatively impact sleep. Greater recognition of these sleep disorders is needed.

An intelligent system for multivariate statistical process monitoring and diagnosis

A knowledge-based system (KBS) was designed for automated system identification, process monitoring, and diagnosis of sensor faults. The real-time KBS consists of a supervisory system using G2 KBS development software linked with external statistical modules for system identification and sensor fault diagnosis. The various statistical techniques were prototyped in MATLAB, converted to ANSI C code, and linked with the G2 Standard Interface. The KBS automatically performs all operations of data collection, identification, monitoring, and sensor fault diagnosis with little or no input from the user. Navigation throughout the KBS is via menu buttons on each user-accessible screen. Selected process variables are displayed on charts showing the history of the variables over a period of time. Multivariate statistical tests and contribution plots are also shown graphically. The KBS was evaluated using simulation studies with a polymerization reactor through a nonlinear dynamic model. Both normal operation conditions as well as conditions of process disturbances were observed to evaluate the KBS performance. Specific user-defined disturbances were added to the simulation, and the KBS correctly diagnosed both process and sensor faults when present.

Use of a novel rule-based expert system in the detection of changes in the ST segment and the T wave in long duration ECGs

The development of a new fast and robust computerised system is examined in detecting electrocardiogram (ECG) changes in long duration ECG recordings. The system distinguishes these changes between ST-segment deviation and T-wave alterations and can support the produced diagnosis by providing explanations for the decisions made. The European Society of Cardiology ST-T Database was used for evaluating the performance of the system. Sensitivity and positive predictive accuracy were the performance measures used and the proposed system scored 92.02% and 93.77%, respectively, in detecting ST-segment episodes and 91.09% and 80.09% in detecting T-wave episodes. By using the chi-square test we also compared the performance of the system between ECG recordings with minimal and substantial amount of noise. The sensitivity of the proposed system is higher than of other algorithms reported in the literature and the positive predictive accuracy is comparable to, or better than, most of them.

Comparability of sleep disorders diagnoses using DSM-IV and ICSD classifications with adolescents

STUDY OBJECTIVES

The use of diagnostic classifications to define sleep disorders is still unusual in epidemiological studies assessing the prevalence of sleep disorders in an adolescent population.

DESIGN

Cross-sectional study. Representative samples of general populations in United Kingdom, Germany and Italy were selected and interviewed by telephone about their sleep habits, sleep and mental disorder diagnoses. Overall, 724 adolescents ages 15-18 years and 1447 young adults ages 19 to 24 years were interviewed. ICSD-90 and DSM-IV diagnoses provided by the Sleep-EVAL expert system were used for the comparisons.

SETTING

N/A.

PARTICIPANTS

N/A.

INTERVENTIONS

N/A.

MEASUREMENTS AND RESULTS

8% of the adolescents and 12.6% of the young adults had ICSD dyssomnia or sleep disturbances associated with a mental disorder. According to the DSM-IV classification, 5.7% of the adolescents and 8.1% of the young adults had a dyssomnia diagnosis. The comparison between the two classifications show that 73.2% of adolescents and young adults with a DSM-IV dyssomnia diagnosis also had similar ICSD diagnosis. The reverse comparison, ICSD vs. DSM-IV, shows that 39.8% of the subjects with an ICSD diagnosis had a DSM-IV diagnosis. DSM-IV primary insomnia was the most frequent diagnosis. Subjects with such a diagnosis were found in about 10 different ICSD diagnoses, mainly inadequate sleep hygiene, psychophysiological or idiopathic insomnia and insufficient sleep syndrome.

CONCLUSIONS

ICSD-90 classification provided higher prevalence of sleep disorder diagnoses than the DSM-IV classification. In adolescents and young adults, DSM-IV primary insomnia is two times more often associated with ICSD inadequate sleep hygiene than with ICSD psychophysiological or idiopathic insomnia.

Clinical workstations supporting evidence-based medicine

The application of the principles of evidence-based medicine is based on a rigorous analysis of clinical research tailored to the individual characteristics of a specific patient. Thus the physician must have information about the patient and about the latest results of clinical research simultaneously. We present a computer-based clinical workstation offering sources for both kinds of information: the current patient data is delivered by access to the electronic patient record and the results of research may be searched at the Internet, Intranet, or other online databases. Performing an evaluation study we are testing this configuration in the setting of a university hospital. The physicians are interviewed about their use of the single information sources and their consequences for clinical research and evidence-based medicine. We hope to show that in spite of some technical and methodological problems the clinical workstation is a promising tool for decision-making at the clinical workplace.

Validation of the sleep-EVAL system against clinical assessments of sleep disorders and polysomnographic data

OBJECTIVES

To validate the Sleep-EVAL expert system, a computerized tool designed for the assessment of sleep disorders, against polysomnographic data and clinical assessments by sleep specialists.

DESIGN

Patients were interviewed twice, once by a physician using Sleep-EVAL and again by a sleep specialist. Polysomnographic data were also recorded to ascertain diagnoses. Agreement between diagnoses generated by Sleep-EVAL and those formulated by sleep specialists was determined via the kappa statistic.

SETTINGS

Sleep disorder centers at Stanford University (USA) and Regensburg University (Germany).

PATIENTS

105 patients aged 18 years or over.

INTERVENTIONS

NA.

RESULTS

Sleep-EVAL made an average of 1.32 diagnoses per patient, compared with 0.93 for the sleep specialists. Overall agreement on any sleep-breathing disorder was 96.9% (Kappa .94). More than half of the patients were diagnosed with obstructive sleep apnea syndrome (OSAS); the agreement rate for this specific diagnosis was 96.7% (Kappa .93).

CONCLUSIONS

The findings indicate that the Sleep-EVAL system is a valid instrument for the recognition of major sleep disorders, particularly insomnia and OSAS.

The EMG diagnosis--an interpretation based on partial information

There is a large difference between the prevalence of a given disease in the general population and in the population seen in the EMG lab. It can be argued that both prevalences are the correct choice as prior probabilities for the diseases. This paradox is resolved by recognizing that the EMG diagnosis is only based on the information provided by the EMG examination and thus only represents a partial view of the patient. We propose a solution summarizing the set of findings, signs and symptoms, lab results etc., that led to the referral of the patient for an EMG examination. This information is described by stochastic variables called FIDL factors (Found In Doctor's Lab). The approach is tested on the EMG expert system MUNIN with 30 previously evaluated cases. The results show that this solution improves the specificity of the diagnosis, without affecting the sensitivity.

Standardizing data collection and decision making with an expert system

The software for the pilot system has been completed. The appropriateness of the risk factor weights needs to be evaluated by clinical testing. However, this does not prevent the system from being used to teach the philosophy of risk group identification and selection of different management strategies according to disease activity. The current system does demonstrate a dynamic relationship between caries risk assessment/activity and different management strategies. A formal scientific evaluation of the effectiveness of the system as a teaching tool is being developed.

Evaluation of an expert system on lymph node pathology

BACKGROUND

Pathfinder is an Expert System that assists pathologists in making accurate diagnoses in the domain of lymph-node pathology. Pathfinder provides a differential diagnosis based on the initial histological feature(s) observed by the pathologist, and suggests to the user additional histological features for observation that are likely to narrow the differential diagnosis.

PURPOSE

To evaluate the diagnostic accuracy of pathologists with and without the Pathfinder.

METHODS

Thirty H&E stained slides from 30 lymph node biopsy specimens on which a Consensus diagnosis was made by two experts were reviewed by 19 pathologists to evaluate Pathfinder. After a period of training, 10 pathologists using Pathfinder (Interactive Computer Method) and 9 pathologists using the Routine Method (diagnosis without computer) determined a differential diagnosis for 15 slides (Test 1). Pathologists were then crossed over, trained, and evaluated the remaining 15 slides (Test 2). For each test, the proportion of "correct" diagnoses was compared between methods. In addition, the information integration attributes (making logical diagnosis given a set of specific histologic features) of Pathfinder and pathologists were compared. Finally, feature identification and quantification skills of pathologists were determined and correlated with the percent correct diagnosis.

RESULTS

The diagnostic accuracy using Pathfinder was greater than that using the Routine Method (40% v 32%, P = .02). Diagnostic accuracy for the group of pathologists who made diagnosis using the Routine Method in Test 1 increased when they made diagnosis using Pathfinder in Test 2 (27% to 44%, P < .0001). The proportion of correct diagnosis for the group of pathologists who used Pathfinder in Test 1 remained virtually unchanged when they gave up Pathfinder in Test 2 (35% to 37%). The percentage of incompatible feature identification ("atypical proliferation" diagnosis) was significantly lower after using Pathfinder (P < .0001). In addition, information integration attributes of Pathfinder were significantly superior than that of the pathologists (P < .0001).

CONCLUSIONS

Pathfinder is a valuable tool that assists pathologists in making accurate diagnosis because it has superior attributes than pathologists to integrate information and to screen for observations incompatible with any specific disease.

Computer modelling of human behaviour in aircraft fire accidents

The mathematical simulation of the evacuation process has a wide and largely untapped scope of application within the aircraft industry. The function of the mathematical model is to provide insight into complex behaviour by allowing designers, legislators, and investigators to ask 'what if' questions. Such a model, EXODUS, is currently under development, and this paper describes its evolution and potential applications. EXODUS is an egress model designed to simulate the evacuation of large numbers of individuals from an enclosure, such as an aircraft. The model tracks the trajectory of each individual as they make their way out of the enclosure or are overcome by fire hazards, such as heat and toxic gases. The software is expert system-based, the progressive motion and behaviour of each individual being determined by a set of heuristics or rules. EXODUS comprises five core interacting components: (i) the Movement Submodel -- controls the physical movement of individual passengers from their current position to the most suitable neighbouring location; (ii) the Behaviour Submodel -- determines an individual's response to the current prevailing situation; (iii) the Passenger Submodel -- describes an individual as a collection of 22 defining attributes and variables; (iv) the Hazard Submodel -- controls the atmospheric and physical environment; and (v) the Toxicity Submodel -- determines the effects on an individual exposed to the fire products, heat, and narcotic gases through the Fractional Effective Dose calculations. These components are briefly described and their capabilities and limitations are demonstrated through comparison with experimental data and several hypothetical evacuation scenarios.

Performance of a neural network trained to make third-molar treatment-planning decisions

The authors developed and tested 12 neural networks of different architectures to make lower-third-molar treatment-planning decisions, using a software-based neural network (Neudesk 1.2, Neural Computer Sciences, Southampton, UK). Network training was undertaken using clinical histories from 119 patients (with 238 lower third molars) referred for treatment planning (79 females and 40 males, mean age 25 years) together with output data consisting of actual treatments planned by a senior oral surgeon. Both the input clinical data and the consultant decisions were treated on a tooth-wise basis and were coded to numerical values. Binary data (e.g., present/absent) were coded to 1 and 0, while quantitative data (e.g., age) were scaled to fall between 0 and 1. A network based on the optimal architecture was trained and then interrogated with test data derived from a further 174 patients (119 females and 55 males, mean age 26 years) with 348 lower third molars. Network decisions were dichotomized with a threshold of 0.8. With no knowledge of the network decisions, the senior oral surgeon indicated his preferred treatments. The teeth were then assigned to "gold-standard" categories of indications present or absent based on National Institutes of Health consensus criteria. Against this, the network achieved a sensitivity of 0.78, which was slightly inferior to that of the oral surgeon (0.88), although this difference was not significant, and a specificity of 0.98, compared with 0.99 for the oral surgeon (p = NS). Agreement between the oral surgeon and network decisions was very high (kappa = 0.850). This study demonstrates that it is possible to train a neural network to provide reliable decision support for lower-third-molar treatment planning.

Multiparameter case studies using knowledge-based systems in hematology

Knowledge-based systems in diagnostic medicine are often used for teaching medical decision making. We have extended the educational value of our suite of hematology decision support systems (peripheral blood analysis, flow cytometry immunophenotyping and DNA analysis, and bone marrow morphology) by creating case studies in a hypertext format. The case studies use the knowledge-based system screens as a background. Digitized images of blood and bone marrow smears from the teaching cases can be accessed on-line. The case studies, which are designed primarily for technologists and physicians, emphasize the proper multiparameter approach to hematopathology diagnosis.

[An expert system for the planning of whole-body irradiation treatments]

Total body irradiation (TBI) combined with intensive chemotherapy and bone marrow transplantation is used with increasing success for the treatment of hematologic malignancies with severe prognosis. An expert system, developed on the basis of the experience acquired during the last years in our department, has been designed to optimize the different treatment steps. The choice of an expert system is based on its capabilities in reconstructing a rule-based reasoning through the combination of theoretical and empirical knowledge. Particularly, a subsystem dealing with the steps to be taken to optimize treatment in TBI has been designed using medical and physical data. Our system can determine radiation treatment variables together with a sequence of quality control procedures. Moreover, indications are given about short, medium and long term damage probabilities and toxicity estimates, both derived from literature data and our personal series of cases. Treatment quality can be assessed and the different techniques compared using these data. This project is aimed at providing physicians and physicists with useful clinical suggestions for TBI setting for bone marrow transplantation.

The development of a decision support system for the pathological diagnosis of human cerebral tumours based on a neural network classifier

This study describes the use of a topological mapping system in the classification of cerebral tumours and the development of a decision support system based upon that classifier. Fourteen pathological parameters from two hundred primary cerebral tumours are presented as vectors to a topological map. The map, consisting of a grid of neurones, learns the features of each tumour by means of a shortest Euclidean distance algorithm, after which self adaptation of the neurons occurs. An LVQ algorithm performs the final classification. Study of the map reveals that it can correctly classify tumors following their malignancy potential and their cytogenesis. The decision support system uses the network at its core and helps not only in reaching a diagnosis but also in finding the optimal way to reach that diagnosis. The usefulness of such a mapping system lies in the field of education, clinical research and medically acceptable cost reduction.

[P.A.I.S., a personal medical information system. A comprehensive medical knowledge base]

The electronic medical knowledge data base DOPIS is a compliation of knowledge from various special fields of medicine. Using uniform nomenclature, the data are presented on demand as they would be in a book chapter. Concise updates can be performed at low cost. The primary structure of the concept is the division of medical knowledge into data banks on diagnosis, literature, medication and pharmacology, as well as so-called electronic textbooks. All data banks and electronic textbooks are connected associatively. Visual information is obtained via the image data bank connected to the diagnosis data bank and the electronic books. Moreover, DOPIS has an integrated patient findings system, as well as an image processing and archiving system with research values enabling research functions. The diagnosis and literature data banks can be modified by the user or author, or fed with their own data (a so-called Expert System Shell). For authors from special fields working on the project, an extra Medical Electronic Publishing System has been developed and made available for the electronic textbooks. The model for the knowledge data base has been developed in the field of ENT, the programme implemented and initially ENT data have been stored.

An expert system for the interpretation of flow cytometric immunophenotyping data

The development of high-grade non-Hodgkin's lymphomas in HIV-positive patients and patients with acquired immune deficiency syndrome (AIDS) is a well known phenomenon. The proper classification of these neoplasms often requires a multiparameter approach, including the interpretation of a large panel of immunologic markers analyzed by flow cytometry. The availability of individuals with the required expertise to properly interpret these marker studies is limited. For this reason, we have designed an expert system to automate the analysis of immunophenotyping panels in both HIV-related and non-HIV-related hematopoietic neoplasms. The expert system, which we call "Professor Fidelio", runs on IBM-compatible computers under Windows 3.0. The system is designed to accept any number of markers studied from a repertoire of 35 markers. Professor Fidelio functions on the basis of heuristic classification of defined diagnostic patterns. Nine specific patterns (Stem Cell, Myeloid and/or Monocytic, Erythroid, Megakaryocytic, Immature B-cell, Immature T-cell, Mature B-cell, Mature T-cell, and Plasma cell) and one "non-specific" pattern have been agreed upon. Fidelio's knowledge base contains the definitions of each of these patterns and the heuristics for excluding patterns when an incomplete panel of markers is performed. The inference engine interprets the findings (including the age of the patient) and reports the patterns which are matched, the differential diagnosis, the suggested diagnosis from the list of differentials if the marker studies are specific, and recommendations for additional tests which may be valuable in establishing the diagnosis.(ABSTRACT TRUNCATED AT 250 WORDS)

A computer assisted toolholder to guide surgery in stereotactic space

A computer assisted toolholder, integrated with an anatomical graphic 3-D rendering programme, is presented. Stereotactic neuroanatomical images are acquired, and the same reference system is employed to represent the position of the toolholder on the monitor. The surgeon can check the orientation of different approach trajectories, moving the toolholder in a situation of virtual reality. Angular values expressed by high precision encoders on the five joints of the toolholder modify "on line" the representation of the configuration of the toolholder within the three dimensional representation of the patient's anatomy.

A multistage system to detect epileptiform activity in the EEG

A PC-based system has been developed to automatically detect epileptiform activity in sixteen-channel bipolar EEG's. The system consists of three stages: data collection, feature extraction, and event detection. The feature extractor employs a mimetic approach to detect candidate epileptiform transients on individual channels, while an expert system is used to detect focal and nonfocal multichannel epileptiform events. Considerable use of spatial and temporal contextual information present in the EEG aids both in the detection of epileptiform events and in the rejection of artifacts and background activity as events. Classification of events as definite or probable overcomes, to some extent, the problem of maintaining high detection rates while eliminating false detections. So far, the system has only been evaluated on development data but, although this does not provide a true measure of performance, the results are nevertheless impressive. Data from 11 patients, totaling 180 minutes of sixteen-channel bipolar EEG's, have been analyzed. A total of 45-71% (average 58%) of epileptiform events reported by the human expert in any EEG were detected as definite with no false detections (i.e., 100% selectivity) and 60-100% (average 80%) as either definite or probable but at the expense of up to nine false detections per hour. Importantly, the highest detection rates were achieved on EEG's containing little epileptiform activity and no false detections were made on normal EEG's.

Knowledge-based design of removable partial dentures using direct manipulation and critiquing

A knowledge-based system for designing removable partial dentures is described in which graphical representations of denture components are manipulated directly by the user to build the required denture design. The arch form of individual patients may be depicted by linear movement and rotation of the icons representing teeth. The size, shape and position of many denture components are determined dynamically to conform to the shape of abutment teeth and to the juxtaposition of other elements. Expert clinical knowledge is used to analyse the developing design and suggest alternate approaches.