Artificial Intelligence-Enhanced Multi-Algorithm R Shiny Application for Predictive Modeling and Analytics: Case...

Area of Science:

• Neurology and Artificial Intelligence
• Biomedical Informatics
• Computational Neuroscience

Background:

• Artificial intelligence (AI) demonstrates superior diagnostic accuracy in healthcare, with growing importance in medical practice.
• SMART-Pred is an innovative AI-based application designed for Alzheimer disease (AD) prediction through handwriting analysis.

Purpose of the Study:

• To develop and evaluate a noninvasive, cost-effective AI tool for early Alzheimer disease (AD) detection.
• To address the need for accessible and accurate screening methods for AD.

Main Methods:

• Utilized principal component analysis for dimensionality reduction of handwriting data.
• Trained and evaluated 10 diverse AI models, including neural networks, on the DARWIN dataset (174 participants).
• Assessed model performance using accuracy, sensitivity, specificity, and AUC metrics, incorporating explainable AI (Shapley Additive Explanations).

Main Results:

• A neural network classifier achieved 91% accuracy and 94% AUC on the test set, surpassing current clinical diagnostic tools.
• "Air_time" and "paper_time" consistently emerged as critical predictors for AD across all models.
• The AI tool's performance aligns with recent advancements in AI-assisted AD prediction.

Conclusions:

• SMART-Pred offers a noninvasive, cost-effective, and efficient method for AD prediction, showcasing AI's potential in healthcare.
• Further clinical validation is needed, but findings support AI-assisted AD diagnosis for improved patient outcomes through early detection.