Multi-label classification of chronically ill patients with bag of words and supervised dimensionality reduction algorithms

Detecting mental and physical disorders using multi-task learning equipped with knowledge graph attention network

Mental and physical disorders (MPD) are inextricably linked in many medical cases; psychosomatic diseases can be induced by mental concerns and psychological discomfort can ensue from physiological diseases. However, existing medical informatics studies focus on identifying mental or physical disorders from a unilateral perspective. Consequently, no existing domain knowledge base, corpus, or detection modeling approach considers mental as well as physical aspects concurrently. This paper proposes a joint modeling approach to detect MPD. First, we crawl through online medical consultation records of patients from websites and build an MPD knowledge ontology by extracting the core conceptual features of the text. Based on the ontology, an MPD knowledge graph containing 12,673 nodes and 82,195 relations is obtained using term matching with a domain thesaurus of each concept. Subsequently, an MPD corpus with fine-grained severities (None, Mild, Moderate, Severe, Dangerous) and 8909 records is constructed by formulating MPD classification criteria and a data annotation process under the guidance of domain experts. Taking the knowledge graph and corpus as the dataset, we design a multi-task learning model to detect the MPD severity, in which a knowledge graph attention network (KGAT) is embedded to better extract knowledge features. Experiments are performed to demonstrate the effectiveness of our model. Furthermore, we employ ontology-based and centrality-based methods to discover additional potential inferred knowledge, which can be captured by KGAT so as to improve the prediction performance and interpretability of our model. Our dataset has been made publicly available, so it can be further used as a medical informatics reference in the fields of psychosomatic medicine, psychiatrics, physical co-morbidity, and so on.

Identification of social determinants of health using multi-label classification of electronic health record clinical notes

OBJECTIVES

Social determinants of health (SDH), key contributors to health, are rarely systematically measured and collected in the electronic health record (EHR). We investigate how to leverage clinical notes using novel applications of multi-label learning (MLL) to classify SDH in mental health and substance use disorder patients who frequent the emergency department.

METHODS AND MATERIALS

We labeled a gold-standard corpus of EHR clinical note sentences (N = 4063) with 6 identified SDH-related domains recommended by the Institute of Medicine for inclusion in the EHR. We then trained 5 classification models: linear-Support Vector Machine, K-Nearest Neighbors, Random Forest, XGBoost, and bidirectional Long Short-Term Memory (BI-LSTM). We adopted 5 common evaluation measures: accuracy, average precision-recall (AP), area under the curve receiver operating characteristic (AUC-ROC), Hamming loss, and log loss to compare the performance of different methods for MLL classification using the F1 score as the primary evaluation metric.

RESULTS

Our results suggested that, overall, BI-LSTM outperformed the other classification models in terms of AUC-ROC (93.9), AP (0.76), and Hamming loss (0.12). The AUC-ROC values of MLL models of SDH related domains varied between (0.59-1.0). We found that 44.6% of our study population (N = 1119) had at least one positive documentation of SDH.

DISCUSSION AND CONCLUSION

The proposed approach of training an MLL model on an SDH rich data source can produce a high performing classifier using only unstructured clinical notes. We also provide evidence that model performance is associated with lexical diversity by health professionals and the auto-generation of clinical note sentences to document SDH.

Clinical decision-making framework against over-testing based on modeling implicit evaluation criteria

Different statistical methods include various subjective criteria that can prevent over-testing. However, no unified framework that defines generalized objective criteria for various diseases is available to determine the appropriateness of diagnostic tests recommended by doctors. We present the clinical decision-making framework against over-testing based on modeling the implicit evaluation criteria (CDFO-MIEC). The CDFO-MIEC quantifies the subjective evaluation process using statistics-based methods to identify over-testing. Furthermore, it determines the test's appropriateness with extracted entities obtained via named entity recognition and entity alignment. More specifically, implicit evaluation criteria are defined-namely, the correlation among the diagnostic tests, symptoms, and diseases, confirmation function, and exclusion function. Additionally, four evaluation strategies are implemented by applying statistical methods, including the multi-label k-nearest neighbor and the conditional probability algorithms, to model the implicit evaluation criteria. Finally, they are combined into a classification and regression tree to make the final decision. The CDFO-MIEC also provides interpretability by decision conditions for supporting each clinical decision of over-testing. We tested the CDFO-MIEC on 2,860 clinical texts obtained from a single respiratory medicine department in China with the appropriate confirmation by physicians. The dataset was supplemented with random inappropriate tests. The proposed framework excelled against the best competing text classification methods with a Mean_F₁ of 0.9167. This determined whether the appropriate and inappropriate tests were properly classified. The four evaluation strategies captured the features effectively, and they were imperative. Therefore, the proposed CDFO-MIEC is feasible because it exhibits high performance and can prevent over-testing.

Multi-label classification methods for improving comorbidities identification

The medical diagnostic process may be supported by computational classification techniques. In many cases, patients are affected by multiple illnesses, and more than one classification label is required to improve medical decision-making. In this paper, we consider a multi-perspective classification problem for medical diagnostics, where cases are described by labels from separate sets. We attempt to improve the identification of comorbidities using multi-label classification techniques. Several investigated methods, which provide label dependencies, are analysed and evaluated. The methods' performances are verified by experiments conducted on four sets of medical data from subject patients. The results were evaluated using several metrics and were statistically verified. We compare the effects of the techniques that do and do not consider label correlations. We demonstrate that multi-label classification methods from the first group outperform the techniques from the second one.

RETRACTED: Diagnosis labeling with disease-specific characteristics mining

This article has been retracted: please see Elsevier Policy on Article Withdrawal (https://www.elsevier.com/about/our-business/policies/article-withdrawal). This article has been retracted at the request of the authors; serious errors had been introduced within the data set presented and, as such, the authors have decided to retract the paper. Error summary is outlined below: 1. The dataset was divided into training set, validation set and testing set. The training set and validation set were used to tune the parameter. According to the authors this is an overfitting and can mislead the readers. 2. The compared baselines are not strong enough for fair comparison. The code was not requested from the original authors, but algorithms have been implemented. 3. The bag of words model is not trained suitably. The model was trained with the used dataset and the performance has been evaluated on this dataset. Authors think this is an overfitting and makes the comparison untrustworthy.

Translational Radiomics: Defining the Strategy Pipeline and Considerations for Application-Part 2: From Clinical Implementation to Enterprise

Enterprise imaging has channeled various technological innovations to the field of clinical radiology, ranging from advanced imaging equipment and postacquisition iterative reconstruction tools to image analysis and computer-aided detection tools. More recently, the advancement in the field of quantitative image analysis coupled with machine learning-based data analytics, classification, and integration has ushered in the era of radiomics, a paradigm shift that holds tremendous potential in clinical decision support as well as drug discovery. However, there are important issues to consider to incorporate radiomics into a clinically applicable system and a commercially viable solution. In this two-part series, we offer insights into the development of the translational pipeline for radiomics from methodology to clinical implementation (Part 1) and from that point to enterprise development (Part 2). In Part 2 of this two-part series, we study the components of the strategy pipeline, from clinical implementation to building enterprise solutions.

Errors, Omissions, and Outliers in Hourly Vital Signs Measurements in Intensive Care

OBJECTIVE

To empirically examine the prevalence of errors, omissions, and outliers in hourly vital signs recorded in the ICU.

DESIGN

Retrospective analysis of vital signs measurements from a large-scale clinical data warehouse (Multiparameter Intelligent Monitoring in Intensive Care III).

SETTING

Data were collected from the medical, surgical, cardiac, and cardiac surgery ICUs of a tertiary medical center in the United States.

PATIENTS

We analyzed data from approximately 48,000 ICU stays including approximately 28 million vital signs measurements.

INTERVENTIONS

None.

MEASUREMENTS AND MAIN RESULTS

We used the vital sign day as our unit of measurement, defined as all the recordings from a single patient for a specific vital sign over a single 24-hour period. Approximately 30-40% of vital sign days included at least one gap of greater than 70 minutes between measurements. Between 3% and 10% of blood pressure measurements included logical inconsistencies. With the exception of pulse oximetry vital sign days, the readings in most vital sign days were normally distributed. We found that 15-38% of vital sign days contained at least one statistical outlier, of which 6-19% occurred simultaneously with outliers in other vital signs.

CONCLUSIONS

We found a significant number of missing, erroneous, and outlying vital signs measurements in a large ICU database. Our results provide empirical evidence of the nonrepresentativeness of hourly vital signs. Additional studies should focus on determining optimal sampling frequencies for recording vital signs in the ICU.

A clinical decision-making mechanism for context-aware and patient-specific remote monitoring systems using the correlations of multiple vital signs

BACKGROUND AND OBJECTIVES

In home-based context-aware monitoring patient's real-time data of multiple vital signs (e.g. heart rate, blood pressure) are continuously generated from wearable sensors. The changes in such vital parameters are highly correlated. They are also patient-centric and can be either recurrent or can fluctuate. The objective of this study is to develop an intelligent method for personalized monitoring and clinical decision support through early estimation of patient-specific vital sign values, and prediction of anomalies using the interrelation among multiple vital signs.

METHODS

In this paper, multi-label classification algorithms are applied in classifier design to forecast these values and related abnormalities. We proposed a completely new approach of patient-specific vital sign prediction system using their correlations. The developed technique can guide healthcare professionals to make accurate clinical decisions. Moreover, our model can support many patients with various clinical conditions concurrently by utilizing the power of cloud computing technology. The developed method also reduces the rate of false predictions in remote monitoring centres.

RESULTS

In the experimental settings, the statistical features and correlations of six vital signs are formulated as multi-label classification problem. Eight multi-label classification algorithms along with three fundamental machine learning algorithms are used and tested on a public dataset of 85 patients. Different multi-label classification evaluation measures such as Hamming score, F1-micro average, and accuracy are used for interpreting the prediction performance of patient-specific situation classifications. We achieved 90-95% Hamming score values across 24 classifier combinations for 85 different patients used in our experiment. The results are compared with single-label classifiers and without considering the correlations among the vitals. The comparisons show that multi-label method is the best technique for this problem domain.

CONCLUSIONS

The evaluation results reveal that multi-label classification techniques using the correlations among multiple vitals are effective ways for early estimation of future values of those vitals. In context-aware remote monitoring this process can greatly help the doctors in quick diagnostic decision making.

Type 2 Diabetes Patients Benefit from the COMODITY12 mHealth System: Results of a Randomised Trial

Patient acceptance is one of the major barriers toward widespread use of mHealth systems. The aim of this study was to assess system operability and whole trial feasibility, including patients' experience with their use of COMMODITY12 mHealth system under. Secondary study aims included assessment of several metabolic parameters as well as patient adherence to the treatment. This was a prospective parallel-arm randomized controlled trial in outpatients diagnosed with DM2, being treated in the primary care settings in Lodz region, Poland, with 6 weeks period of follow-up. Patients opinions were collected with 7-item questionnaire, assessing different aspects of system use, as well as EuroQol-5D-5 L questionnaire, assessing health-related quality of life. Sixty patients (female, 24, male, 36, mean age +/- SD 59.5 +/- 6.8) completed study. All four layers of the COMMODITY12 system proved to work smooth under real-life conditions, without major problems. All dimensions of experience with system use were assessed well, with maximum values for clearness of instructions, and ease of use (4.80, and 4.63, respectively). Health related quality of life, as assessed with cumulative utility measure, improved significantly in COMMODITY12 system users (P < 0.05). mHealth system modestly improved glycaemic and blood pressure control, assuring high level of patient adherence with overall adherence reaching 92.9 %. Study proved that the COMODITY12 system is well accepted by type 2 diabetes patients taking part in clinical trial, leading to several clinical benefits, and improved quality of life. Nevertheless, before future commercialisation of the system, several minor problems identified during the study need to be addressed.

Performance comparison of multi-label learning algorithms on clinical data for chronic diseases

We are motivated by the issue of classifying diseases of chronically ill patients to assist physicians in their everyday work. Our goal is to provide a performance comparison of state-of-the-art multi-label learning algorithms for the analysis of multivariate sequential clinical data from medical records of patients affected by chronic diseases. As a matter of fact, the multi-label learning approach appears to be a good candidate for modeling overlapped medical conditions, specific to chronically ill patients. With the availability of such comparison study, the evaluation of new algorithms should be enhanced. According to the method, we choose a summary statistics approach for the processing of the sequential clinical data, so that the extracted features maintain an interpretable link to their corresponding medical records. The publicly available MIMIC-II dataset, which contains more than 19,000 patients with chronic diseases, is used in this study. For the comparison we selected the following multi-label algorithms: ML-kNN, AdaBoostMH, binary relevance, classifier chains, HOMER and RAkEL. Regarding the results, binary relevance approaches, despite their elementary design and their independence assumption concerning the chronic illnesses, perform optimally in most scenarios, in particular for the detection of relevant diseases. In addition, binary relevance approaches scale up to large dataset and are easy to learn. However, the RAkEL algorithm, despite its scalability problems when it is confronted to large dataset, performs well in the scenario which consists of the ranking of the labels according to the dominant disease of the patient.