Ethical issues when using digital biomarkers and artificial intelligence for the early detection of dementia

Machine learning-based screening for outpatients with dementia using drawing features from the clock drawing test

Background and Objectives: In geriatrics and dementia care, early diagnosis is crucial. We developed a dementia screening model using drawing features from clock drawing tests (CDT) and investigated the features contributing to the discrimination of dementia and its screening performance. Methods: This study included 129 older adults attending a dementia outpatient clinic. We obtained information on the diagnosis of dementia and CDT data from medical records and quantified 12 types of drawing features according to the Freedman scoring system. Based on the dementia diagnosis information, participants were assigned to two groups: 58 in the dementia diagnosis group and 71 in the non-diagnosis group. Using Boruta, an iterative feature selection algorithm, and a support vector machine, a machine learning method, we analyzed the drawing features contributing to dementia discrimination and evaluated discrimination performance. Results: Five types of drawing features were selected as contributors to discrimination, including "numbers in the correct position," "minute target number indicated," and "hand in correct proportion." These features exhibited a discriminating sensitivity of 0.74 ± 0.16 and specificity of 0.74 ± 0.18 for detecting dementia. Conclusion: This study demonstrated a method for identifying individuals likely to be diagnosed with dementia among patients attending a dementia outpatient clinic using drawing features. The knowledge of drawing features contributing to dementia differentiation may assist healthcare practitioners in clinical reasoning and provide novel insights for clinical practice. In the future, we plan to develop a primary screening for dementia based on machine learning using CDT.

Identifying individuals at high risk for dementia in primary care: Development and validation of the DemRisk risk prediction model using routinely collected patient data

INTRODUCTION

Health policy in the UK and globally regarding dementia, emphasises prevention and risk reduction. These goals could be facilitated by automated assessment of dementia risk in primary care using routinely collected patient data. However, existing applicable tools are weak at identifying patients at high risk for dementia. We set out to develop improved risk prediction models deployable in primary care.

METHODS

Electronic health records (EHRs) for patients aged 60-89 from 393 English general practices were extracted from the Clinical Practice Research Datalink (CPRD) GOLD database. 235 and 158 practices respectively were randomly assigned to development and validation cohorts. Separate dementia risk models were developed for patients aged 60-79 (development cohort n = 616,366; validation cohort n = 419,126) and 80-89 (n = 175,131 and n = 118,717). The outcome was incident dementia within 5 years and more than 60 evidence-based risk factors were evaluated. Risk models were developed and validated using multivariable Cox regression.

RESULTS

The age 60-79 development cohort included 10,841 incident cases of dementia (6.3 per 1,000 person-years) and the age 80-89 development cohort included 15,994 (40.2 per 1,000 person-years). Discrimination and calibration for the resulting age 60-79 model were good (Harrell's C 0.78 (95% CI: 0.78 to 0.79); Royston's D 1.74 (1.70 to 1.78); calibration slope 0.98 (0.96 to 1.01)), with 37% of patients in the top 1% of risk scores receiving a dementia diagnosis within 5 years. Fit statistics were lower for the age 80-89 model but dementia incidence was higher and 79% of those in the top 1% of risk scores subsequently developed dementia.

CONCLUSION

Our models can identify individuals at higher risk of dementia using routinely collected information from their primary care record, and outperform an existing EHR-based tool. Discriminative ability was greatest for those aged 60-79, but the model for those aged 80-89 may also be clinical useful.

Dementia risk prediction using decision-focused content selection from medical notes

Several general-purpose language model (LM) architectures have been proposed with demonstrated improvement in text summarization and classification. Adapting these architectures to the medical domain requires additional considerations. For instance, the medical history of the patient is documented in the Electronic Health Record (EHR) which includes many medical notes drafted by healthcare providers. Direct processing of these notes may not be possible because the computational complexity of LMs imposes a limit on the length of input text. Therefore, previous applications resorted to content selection using truncation or summarization of the text. Unfortunately, these text processing techniques may lead to information loss, redundancy or irrelevance. In the present paper, a decision-focused content selection technique is proposed. The objective of this technique is to select a subset of sentences from the medical notes of a patient that are relevant to the target outcome over a predefined observation period. This decision-focused content selection methodology is then used to develop a dementia risk prediction model based on the Longformer LM architecture. The results show that the proposed framework delivers an AUC of 78.43 when the summary is restricted to 1024 tokens, outperforming previously proposed content selection techniques. This performance is notable given that the model estimates dementia risk with a one year prediction horizon, relies on an observation period of only one year and solely uses medical notes without other EHR data modalities. Moreover, the proposed techniques overcome the limitation of machine learning models that use a tabular representation of the text by preserving contextual content, enable feature engineering from raw text and circumvent the computational complexity of language models.

Smartphone-Based Hand Function Assessment: Systematic Review

BACKGROUND

Hand function assessment heavily relies on specific task scenarios, making it challenging to ensure validity and reliability. In addition, the wide range of assessment tools, limited and expensive data recording, and analysis systems further aggravate the issue. However, smartphones provide a promising opportunity to address these challenges. Thus, the built-in, high-efficiency sensors in smartphones can be used as effective tools for hand function assessment.

OBJECTIVE

This review aims to evaluate existing studies on hand function evaluation using smartphones.

METHODS

An information specialist searched 8 databases on June 8, 2023. The search criteria included two major concepts: (1) smartphone or mobile phone or mHealth and (2) hand function or function assessment. Searches were limited to human studies in the English language and excluded conference proceedings and trial register records. Two reviewers independently screened all studies, with a third reviewer involved in resolving discrepancies. The included studies were rated according to the Mixed Methods Appraisal Tool. One reviewer extracted data on publication, demographics, hand function types, sensors used for hand function assessment, and statistical or machine learning (ML) methods. Accuracy was checked by another reviewer. The data were synthesized and tabulated based on each of the research questions.

RESULTS

In total, 46 studies were included. Overall, 11 types of hand dysfunction-related problems were identified, such as Parkinson disease, wrist injury, stroke, and hand injury, and 6 types of hand dysfunctions were found, namely an abnormal range of motion, tremors, bradykinesia, the decline of fine motor skills, hypokinesia, and nonspecific dysfunction related to hand arthritis. Among all built-in smartphone sensors, the accelerometer was the most used, followed by the smartphone camera. Most studies used statistical methods for data processing, whereas ML algorithms were applied for disease detection, disease severity evaluation, disease prediction, and feature aggregation.

CONCLUSIONS

This systematic review highlights the potential of smartphone-based hand function assessment. The review suggests that a smartphone is a promising tool for hand function evaluation. ML is a conducive method to classify levels of hand dysfunction. Future research could (1) explore a gold standard for smartphone-based hand function assessment and (2) take advantage of smartphones' multiple built-in sensors to assess hand function comprehensively, focus on developing ML methods for processing collected smartphone data, and focus on real-time assessment during rehabilitation training. The limitations of the research are 2-fold. First, the nascent nature of smartphone-based hand function assessment led to limited relevant literature, affecting the evidence's completeness and comprehensiveness. This can hinder supporting viewpoints and drawing conclusions. Second, literature quality varies due to the exploratory nature of the topic, with potential inconsistencies and a lack of high-quality reference studies and meta-analyses.

Ethics of Wearable-Based Out-of-Hospital Cardiac Arrest Detection

Out-of-hospital cardiac arrest is a major health problem, and immediate treatment is essential for improving the chances of survival. The development of technological solutions to detect out-of-hospital cardiac arrest and alert emergency responders is gaining momentum; multiple research consortia are currently developing wearable technology for this purpose. For the responsible design and implementation of this technology, it is necessary to attend to the ethical implications. This review identifies relevant ethical aspects of wearable-based out-of-hospital cardiac arrest detection according to four key principles of medical ethics. First, aspects related to beneficence concern the effectiveness of the technology. Second, nonmaleficence requires preventing psychological distress associated with wearing the device and raises questions about the desirability of screening. Third, grounded in autonomy are empowerment, the potential reidentification from continuously collected data, issues of data access, bystander privacy, and informed consent. Finally, justice concerns include the risks of algorithmic bias and unequal technology access. Based on this overview and relevant legislation, we formulate design recommendations. We suggest that key elements are device accuracy and reliability, dynamic consent, purpose limitation, and personalization. Further empirical research is needed into the perspectives of stakeholders, including people at risk of out-of-hospital cardiac arrest and their next-of-kin, to achieve a successful and ethically balanced integration of this technology in society.

Feasibility of Observing Cerebrovascular Disease Phenotypes with Smartphone Monitoring: Study Design Considerations for Real-World Studies

Accelerated by the adoption of remote monitoring during the COVID-19 pandemic, interest in using digitally captured behavioral data to predict patient outcomes has grown; however, it is unclear how feasible digital phenotyping studies may be in patients with recent ischemic stroke or transient ischemic attack. In this perspective, we present participant feedback and relevant smartphone data metrics suggesting that digital phenotyping of post-stroke depression is feasible. Additionally, we proffer thoughtful considerations for designing feasible real-world study protocols tracking cerebrovascular dysfunction with smartphone sensors.

Mapping the ethical landscape of digital biomarkers: A scoping review

In the evolving landscape of digital medicine, digital biomarkers have emerged as a transformative source of health data, positioning them as an indispensable element for the future of the discipline. This necessitates a comprehensive exploration of the ethical complexities and challenges intrinsic to this cutting-edge technology. To address this imperative, we conducted a scoping review, seeking to distill the scientific literature exploring the ethical dimensions of the use of digital biomarkers. By closely scrutinizing the literature, this review aims to bring to light the underlying ethical issues associated with the development and integration of digital biomarkers into medical practice.

Machine Learning and Digital Biomarkers Can Detect Early Stages of Neurodegenerative Diseases

Neurodegenerative diseases (NDs) such as Alzheimer's Disease (AD) and Parkinson's Disease (PD) are devastating conditions that can develop without noticeable symptoms, causing irreversible damage to neurons before any signs become clinically evident. NDs are a major cause of disability and mortality worldwide. Currently, there are no cures or treatments to halt their progression. Therefore, the development of early detection methods is urgently needed to delay neuronal loss as soon as possible. Despite advancements in Medtech, the early diagnosis of NDs remains a challenge at the intersection of medical, IT, and regulatory fields. Thus, this review explores "digital biomarkers" (tools designed for remote neurocognitive data collection and AI analysis) as a potential solution. The review summarizes that recent studies combining AI with digital biomarkers suggest the possibility of identifying pre-symptomatic indicators of NDs. For instance, research utilizing convolutional neural networks for eye tracking has achieved significant diagnostic accuracies. ROC-AUC scores reached up to 0.88, indicating high model performance in differentiating between PD patients and healthy controls. Similarly, advancements in facial expression analysis through tools have demonstrated significant potential in detecting emotional changes in ND patients, with some models reaching an accuracy of 0.89 and a precision of 0.85. This review follows a structured approach to article selection, starting with a comprehensive database search and culminating in a rigorous quality assessment and meaning for NDs of the different methods. The process is visualized in 10 tables with 54 parameters describing different approaches and their consequences for understanding various mechanisms in ND changes. However, these methods also face challenges related to data accuracy and privacy concerns. To address these issues, this review proposes strategies that emphasize the need for rigorous validation and rapid integration into clinical practice. Such integration could transform ND diagnostics, making early detection tools more cost-effective and globally accessible. In conclusion, this review underscores the urgent need to incorporate validated digital health tools into mainstream medical practice. This integration could indicate a new era in the early diagnosis of neurodegenerative diseases, potentially altering the trajectory of these conditions for millions worldwide. Thus, by highlighting specific and statistically significant findings, this review demonstrates the current progress in this field and the potential impact of these advancements on the global management of NDs.

Remote data collection speech analysis in people at risk for Alzheimer's disease dementia: usability and acceptability results

Introduction

Digital cognitive assessments are gathering importance for the decentralized remote clinical trials of the future. Before including such assessments in clinical trials, they must be tested to confirm feasibility and acceptability with the intended participant group. This study presents usability and acceptability data from the Speech on the Phone Assessment (SPeAk) study.

Methods

Participants (N = 68, mean age 70.43 years, 52.9% male) provided demographic data and completed baseline and 3-month follow-up phone based assessments. The baseline visit was administered by a trained researcher and included a spontaneous speech assessment and a brief cognitive battery (immediate and delayed recall, digit span, and verbal fluency). The follow-up visit repeated the cognitive battery which was administered by an automatic phone bot. Participants were randomized to receive their cognitive test results acer the final or acer each study visit. Participants completed acceptability questionnaires electronically acer each study visit.

Results

There was excellent retention (98.5%), few technical issues (n = 5), and good interrater reliability. Participants rated the assessment as acceptable, confirming the ease of use of the technology and their comfort in completing cognitive tasks on the phone. Participants generally reported feeling happy to receive the results of their cognitive tests, and this disclosure did not cause participants to feel worried.

Discussion

The results from this usability and acceptability analysis suggest that completing this brief battery of cognitive tests via a telephone call is both acceptable and feasible in a midlife-to-older adult population in the United Kingdom, living at risk for Alzheimer's disease.

Artificial Intelligence in Dementia: A Bibliometric Study

The applications of artificial intelligence (AI) in dementia research have garnered significant attention, prompting the planning of various research endeavors in current and future studies. The objective of this study is to provide a comprehensive overview of the research landscape regarding AI and dementia within scholarly publications and to suggest further studies for this emerging research field. A search was conducted in the Web of Science database to collect all relevant and highly cited articles on AI-related dementia research published in English until 16 May 2023. Utilizing bibliometric indicators, a search strategy was developed to assess the eligibility of titles, utilizing abstracts and full texts as necessary. The Bibliometrix tool, a statistical package in R, was used to produce and visualize networks depicting the co-occurrence of authors, research institutions, countries, citations, and keywords. We obtained a total of 1094 relevant articles published between 1997 and 2023. The number of annual publications demonstrated an increasing trend over the past 27 years. Journal of Alzheimer's Disease (39/1094, 3.56%), Frontiers in Aging Neuroscience (38/1094, 3.47%), and Scientific Reports (26/1094, 2.37%) were the most common journals for this domain. The United States (283/1094, 25.86%), China (222/1094, 20.29%), India (150/1094, 13.71%), and England (96/1094, 8.77%) were the most productive countries of origin. In terms of institutions, Boston University, Columbia University, and the University of Granada demonstrated the highest productivity. As for author contributions, Gorriz JM, Ramirez J, and Salas-Gonzalez D were the most active researchers. While the initial period saw a relatively low number of articles focusing on AI applications for dementia, there has been a noticeable upsurge in research within this domain in recent years (2018-2023). The present analysis sheds light on the key contributors in terms of researchers, institutions, countries, and trending topics that have propelled the advancement of AI in dementia research. These findings collectively underscore that the integration of AI with conventional treatment approaches enhances the effectiveness of dementia diagnosis, prediction, classification, and monitoring of treatment progress.