New Horizons in artificial intelligence in the healthcare of older people

Advancing Remote Monitoring for Patients With Alzheimer Disease and Related Dementias: Systematic Review

BACKGROUND

Using remote monitoring technology in the context of Alzheimer disease (AD) care presents exciting new opportunities to lessen caregiver stress and improve patient care quality. The application of wearables, environmental sensors, and smart home systems designed specifically for patients with AD represents a promising interdisciplinary approach that integrates advanced technology with health care to enhance patient safety, monitor health parameters in real time, and provide comprehensive support to caregivers.

OBJECTIVE

The objectives of this study included evaluating the effectiveness of various remote sensing technologies in enhancing patient outcomes and identifying strategies to alleviate the burden on health care professionals and caregivers. Critical elements such as regulatory compliance, user-centered design, privacy and security considerations, and the overall efficacy of relevant technologies were comprehensively examined. Ultimately, this study aimed to propose a comprehensive remote monitoring framework tailored to the needs of patients with AD and related dementias.

METHODS

Guided by the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework, we conducted a systematic review on remote monitoring for patients with AD and related dementias. Our search spanned 4 major electronic databases-Google Scholar, PubMed, IEEE Xplore, and DBLP on February 20, 2024, with an updated search on May 18, 2024.

RESULTS

A total of 31 publications met the inclusion criteria, highlighting 4 key research areas: existing remote monitoring technologies, balancing practicality and empathy, security and privacy in monitoring, and technology design for AD care. The studies revealed a strong focus on various remote monitoring methods for capturing behavioral, physiological, and environmental data yet showed a gap in evaluating these methods for patient and caregiver needs, privacy, and usability. The findings also indicated that many studies lacked robust reference standards and did not consistently apply critical appraisal criteria, underlining the need for comprehensive frameworks that better integrate these essential considerations.

CONCLUSIONS

This comprehensive literature review of remote monitoring technologies for patients with AD provides an understanding of remote monitoring technologies, trends, and gaps in the current research and the significance of novel strategies for remote monitoring to enhance patient outcomes and reduce the burden among health professionals and caregivers. The proposed remote monitoring framework aims to inspire the development of new interdisciplinary research models that advance care for patients with AD.

Around the EQUATOR With Clin-STAR: AI-Based Randomized Controlled Trial Challenges and Opportunities in Aging Research

The CONSORT 2010 statement is a guideline that provides an evidence-based checklist of minimum reporting standards for randomized trials. With the rapid growth of Artificial Intelligence (AI) based interventions in the past 10 years, the CONSORT-AI extension was created in 2020 to provide guidelines for AI-based randomized controlled trials (RCT). The Clin-STAR "Around the EQUATOR" series features existing reported standards while also highlighting the inherent complexities of research involving research of older participants. In this work, we propose that when designing AI-based RCTs involving older adults, researchers adopt a conceptual framework (CONSORT-AI-5Ms) designed around the 5Ms (Mind, Mobility, Medications, Matters most, and Multi-complexity) of Age-Friendly Healthcare Systems. Employing the 5Ms in this context, we provide a detailed rationale and include specific examples of challenges and potential solutions to maximize the impact and value of AI RCTs in an older adult population. By combining the original intent of CONSORT-AI with the 5Ms framework, CONSORT-AI-5Ms provides a patient-centered and equitable perspective to consider when designing AI-based RCTs to address the diverse needs and challenges associated with geriatric care.

An artificial intelligence perspective on geriatric syndromes: assessing the information accuracy and readability of ChatGPT

PURPOSE

ChatGPT, a comprehensive language processing model, provides the opportunity for supportive and professional interactions with patients. However, its use to address patients' frequently asked questions (FAQs) and the readability of the text generated by ChatGPT remain unexplored, particularly in geriatrics. We identified the FAQs about common geriatric syndromes and assessed the accuracy and readability of the responses provided by ChatGPT.

METHODS

Two geriatricians with extensive knowledge and experience in geriatric syndromes independently reviewed the 28 responses provided by ChatGPT. The accuracy of the responses generated by ChatGPT was categorized on a rating scale from 0 (harmful) to 4 (excellent) based on current guidelines and approaches. The readability of the text generated by ChatGPT was assessed by administering two tests: the Flesch-Kincaid Reading Ease (FKRE) and the Flesch-Kincaid Grade Level (FKGL).

RESULTS

ChatGPT-generated responses with an overall mean accuracy score of 88% (3.52/4). Responses generated for sarcopenia diagnosis and depression treatment in older adults had the lowest accuracy scores (2.0 and 2.5, respectively). The mean FKRE score of the texts was 25.2, while the mean FKGL score was 14.5.

CONCLUSION

The accuracy scores of the responses generated by ChatGPT were high in most common geriatric syndromes except for sarcopenia diagnosis and depression treatment. Moreover, the text generated by ChatGPT was very difficult to read and best understood by college graduates. ChatGPT may reduce the uncertainty many patients face. Nevertheless, it remains advisable to consult with subject matter experts when undertaking consequential decision-making.

Perspectives on AI and Novel Technologies Among Older Adults, Clinicians, Payers, Investors, and Developers

Importance

Artificial intelligence (AI) and novel technologies, such as remote sensors, robotics, and decision support algorithms, offer the potential for improving the health and well-being of older adults, but the priorities of key partners across the technology innovation continuum are not well understood.

Objective

To examine the priorities and suggested applications for AI and novel technologies for older adults among key partners.

Design, Setting, and Participants

This qualitative study comprised individual interviews using grounded theory conducted from May 24, 2023, to January 24, 2024. Recruitment occurred via referrals through the Johns Hopkins Artificial Intelligence and Technology Collaboratory for Aging Research. Participants included adults aged 60 years or older or their caregivers, clinicians, leaders in health systems or insurance plans (ie, payers), investors, and technology developers.

Main Outcomes and Measures

To assess priority areas, older adults, caregivers, clinicians, and payers were asked about the most important challenges faced by older adults and their caregivers, and investors and technology developers were asked about the most important opportunities associated with older adults and technology. All participants were asked for suggestions regarding AI and technology applications. Payers, investors, and technology developers were asked about end user engagement, and all groups except technology developers were asked about suggestions for technology development. Interviews were analyzed using qualitative thematic analysis. Distinct priority areas were identified, and the frequency and type of priority areas were compared by participant groups to assess the extent of overlap in priorities across groups.

Results

Participants included 15 older adults or caregivers (mean age, 71.3 years [range, 65-93 years]; 4 men [26.7%]), 15 clinicians (mean age, 50.3 years [range, 33-69 years]; 8 men [53.3%]), 8 payers (mean age, 51.6 years [range, 36-65 years]; 5 men [62.5%]), 5 investors (mean age, 42.4 years [range, 31-56 years]; 5 men [100%]), and 6 technology developers (mean age, 42.0 years [range, 27-62 years]; 6 men [100%]). There were different priorities across key partners, with the most overlap between older adults or caregivers and clinicians and the least overlap between older adults or caregivers and investors and technology developers. Participants suggested novel applications, such as using reminders for motivating self-care or social engagement. There were few to no suggestions that addressed activities of daily living, which was the most frequently reported priority for older adults or caregivers. Although all participants agreed on the importance of engaging end users, engagement challenges included regulatory barriers and stronger influence of payers relative to other end users.

Conclusions and Relevance

This qualitative interview study found important differences in priorities for AI and novel technologies for older adults across key partners. Public health, regulatory, and advocacy strategies are needed to raise awareness about these priorities, foster engagement, and align incentives to effectively use AI to improve the health of older adults.

Developing a National Study That Integrates the Geriatric Assessment into the Care of Older Patients with Myeloid Malignancies

PURPOSE OF REVIEW

To highlight the priorities in geriatric assessment research in myeloid malignancies and discuss design considerations necessary to ensure research is patient-centric, generalizeable, and high quality.

RECENT FINDINGS

Older adults with myeloid malignancies including those who are perceived to have excellent performance status have multiple functional impairments. These impairments are associated with early mortality. Older adults have different functional trajectories through the course of treatment; this will be further investigated in our ongoing multicenter study. In a single-center study, we have demonstrated the use of geriatric assessment to guide treatment is feasible. Key priorities include designing a multicenter validation study to confirm the role of geriatric assessment in determining treatment tolerance and survival. Such a study should include core geriatric assessment measures and should enroll diverse patient population across various practices. Conducting such a study is necessary to advance patient care and trial design, and to open venues to conduct studies to confirm the role of geriatric assessment in treatment selection.

Delirium detection in the emergency department: A diagnostic accuracy meta-analysis of history, physical examination, laboratory tests, and screening instruments

INTRODUCTION

Geriatric emergency department (ED) guidelines emphasize timely identification of delirium. This article updates previous diagnostic accuracy systematic reviews of history, physical examination, laboratory testing, and ED screening instruments for the diagnosis of delirium as well as test-treatment thresholds for ED delirium screening.

METHODS

We conducted a systematic review to quantify the diagnostic accuracy of approaches to identify delirium. Studies were included if they described adults aged 60 or older evaluated in the ED setting with an index test for delirium compared with an acceptable criterion standard for delirium. Data were extracted and studies were reviewed for risk of bias. When appropriate, we conducted a meta-analysis and estimated delirium screening thresholds.

RESULTS

Full-text review was performed on 55 studies and 27 were included in the current analysis. No studies were identified exploring the accuracy of findings on history or laboratory analysis. While two studies reported clinicians accurately rule in delirium, clinician gestalt is inadequate to rule out delirium. We report meta-analysis on three studies that quantified the accuracy of the 4 A's Test (4AT) to rule in (pooled positive likelihood ratio [LR+] 7.5, 95% confidence interval [CI] 2.7-20.7) and rule out (pooled negative likelihood ratio [LR-] 0.18, 95% CI 0.09-0.34) delirium. We also conducted meta-analysis of two studies that quantified the accuracy of the Abbreviated Mental Test-4 (AMT-4) and found that the pooled LR+ (4.3, 95% CI 2.4-7.8) was lower than that observed for the 4AT, but the pooled LR- (0.22, 95% CI 0.05-1) was similar. Based on one study the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) is the superior instrument to rule in delirium. The calculated test threshold is 2% and the treatment threshold is 11%.

CONCLUSIONS

The quantitative accuracy of history and physical examination to identify ED delirium is virtually unexplored. The 4AT has the largest quantity of ED-based research. Other screening instruments may more accurately rule in or rule out delirium. If the goal is to rule in delirium then the CAM-ICU or brief CAM or modified CAM for the ED are superior instruments, although the accuracy of these screening tools are based on single-center studies. To rule out delirium, the Delirium Triage Screen is superior based on one single-center study.

New horizons in prediction modelling using machine learning in older people's healthcare research

Machine learning (ML) and prediction modelling have become increasingly influential in healthcare, providing critical insights and supporting clinical decisions, particularly in the age of big data. This paper serves as an introductory guide for health researchers and readers interested in prediction modelling and explores how these technologies support clinical decisions, particularly with big data, and covers all aspects of the development, assessment and reporting of a model using ML. The paper starts with the importance of prediction modelling for precision medicine. It outlines different types of prediction and machine learning approaches, including supervised, unsupervised and semi-supervised learning, and provides an overview of popular algorithms for various outcomes and settings. It also introduces key theoretical ML concepts. The importance of data quality, preprocessing and unbiased model performance evaluation is highlighted. Concepts of apparent, internal and external validation will be introduced along with metrics for discrimination and calibration for different types of outcomes. Additionally, the paper addresses model interpretation, fairness and implementation in clinical practice. Finally, the paper provides recommendations for reporting and identifies common pitfalls in prediction modelling and machine learning. The aim of the paper is to help readers understand and critically evaluate research papers that present ML models and to serve as a first guide for developing, assessing and implementing their own.

Development and External Validation of a Machine Learning-based Fall Prediction Model for Nursing Home Residents: A Prospective Cohort Study

OBJECTIVES

To develop and externally validate a machine learning-based fall prediction model for ambulatory nursing home residents. The focus is on predicting fall occurrences within 6 months after baseline assessment through a binary classification task, aiming to provide staff with an effective and user-friendly fall-risk assessment tool.

DESIGN

Prospective cohort study.

SETTING AND PARTICIPANTS

A total of 864 older residents living in 4 nursing homes between May 2022 and March 2023 in China.

METHODS

Potential fall-risk predictors were collected through in-person interviews and assessments of anthropometric and physical function. Participants were followed for 6 months, with falls recorded by trained nurses. Seven machine learning algorithms, including Logistic Regression (LR), Gradient Boosting Machine (GBM), eXtreme Gradient Boosting (XGBoost), Random Forest (RF), Support Vector Machine (SVM), Neural Networks (NN), and Decision Tree (DT), were used to develop prediction models. Performance was evaluated using the area under the receiver operating characteristic curve (ROC-AUC) and Precision-Recall curve (PR-AUC), with calibration assessed via a calibration curve. Feature importance was visualized using SHapley Additive exPlanations (SHAP).

RESULTS

The 6 selected predictors were balance, grip strength, fatigue, fall history, age, and comorbidity. The ROC-AUC for the models ranged from 0.710 to 0.750, PR-AUC from 0.415 to 0.473, sensitivity from 0.704 to 0.914, and specificity from 0.511 to 0.687 in the validation cohort. The LR model was converted into a nomogram.

CONCLUSIONS AND IMPLICATIONS

The machine learning-based fall-prediction models effectively identified nursing home residents at high risk of falls. The developed nomogram can be integrated into clinical practice to enhance fall risk assessment protocols, ultimately improving patient safety and care in nursing homes.

Machine learning in female urinary incontinence: A scoping review

Introduction and Hypothesis

The aim was to conduct a scoping review of the literature on the use of machine learning (ML) in female urinary incontinence (UI) over the last decade.

Methods

A systematic search was performed among the Medline, Google Scholar, PubMed, and Web of Science databases using the following keywords: [Urinary incontinence] and [(Machine learning) or (Predict) or (Prediction model)]. Eligible studies were considered to have applied ML model to explore different management processes of female UI. Data analyzed included the field of application, type of ML, input variables, and results of model validation.

Results

A total of 798 papers were identified while 23 finally met the inclusion criteria. The vast majority of studies applied logistic regression to establish models (91.3%, 21/23). Most frequently ML was applied to predict postpartum UI (39.1%, 9/23), followed by de novo incontinence after pelvic floor surgery (34.8%, 8/23).There are also three papers using ML models to predict treatment outcomes and three papers using ML models to assist in diagnosis. Variables for modeling included demographic characteristics, clinical data, pelvic floor ultrasound, and urodynamic parameters. The area under receiver operating characteristic curve of these models fluctuated from 0.56 to 0.95, and only 11 studies reported sensitivity and specificity, with sensitivity ranging from 20% to 96.2% and specificity from 59.8% to 94.5%.

Conclusion

Machine learning modeling demonstrated good predictive and diagnostic abilities in some aspects of female UI, showing its promising prospects in near future. However, the lack of standardization and transparency in the validation and evaluation of the models, and the insufficient external validation greatly diminished the applicability and reproducibility, thus a focus on filling this gap is strongly recommended for future research.