Interpretable machine learning for dementia: A systematic review

Abstract

Introduction:Machine learning research into automated dementia diagnosis is becoming increasingly popular but so far has had limited clinical impact. A key challenge is building robust and generalizable models that generate decisions that can be reliably explained. Some models are designed to be inherently "interpretable," whereas post hoc "explainability" methods can be used for other models.Methods: Here we sought to summarize the state-of-the-art of interpretable machine learning for dementia. Results:We identified 92 studies using PubMed, Web of Science, and Scopus. Studies demonstrate promising classification performance but vary in their validation procedures and reporting standards and rely heavily on popular data sets.Discussion: Future work should incorporate clinicians to validate explanation methods and make conclusive inferences about dementia-related disease pathology. Critically analyzing model explanations also requires an understanding of the interpretability methods itself. Patient-specific explanations are also required to demonstrate the benefit of interpretable machine learning in clinical practice.

Deep‐learning model for prenatal congenital heart disease screening generalizes to community setting and outperforms clinical detection

“…a heatmap that does not focus clearly on clinically relevant features does not necessarily mean that the model's performance is poor. How best to analyze GradCAMs to understand model function is still an active area of research and is beyond the scope of this work [23][24][25] . Nevertheless, one might imagine that a more robustly trained model in the future may yield even better diagnostic performance.…”

Recent advancement in integrating artificial intelligence and information technology with real‐world data for clinical decision‐making in China: A scoping review

“…Although clear benefits on predictive ability have been demonstrated by the AI‐enabled models in comparison with conventional models, 19,82 the application of AI incorporated into RWD to support CDM is still under early development in China. In comparison, though the global trend has seen a similar rise in the popularity of technology‐/AI‐based CDSSs, a more extensive range of therapeutic areas has been covered, including orthopedics, 83 psychiatry and mental health, 84,85 allergy and immunology, 86 anesthesiology, 87 sleep medicine, 88 etc. To further promote clinical application and impact, healthcare providers and research communities worldwide have been actively investigating ways to incorporate AI‐/technology‐based CDSSs into clinical workflows by exploring barriers and facilitators in successful implementation and formulating collaborative strategies 89–91 .…”

Deep learning applications in vascular dementia using neuroimaging

“…Although deep learning techniques continue to redefine the landscape of VaD research, these methods still lack transparency, and the black-box nature of deep learning models make outputs hard to interpret [65]. In medical field, building robust, generalizable, and interpretable models is necessary but challenging [66,67 ▪ ]. Moreover, maintaining patient privacy and data security while utilizing sensitive data is also paramount.…”