Using Artificial Intelligence to Learn Optimal Regimen Plan for Alzheimer's Disease.

Area of Science:

• Neurology
• Artificial Intelligence
• Medical Informatics

Background:

• Alzheimer's Disease (AD) management requires complex clinical decisions due to the lack of curative treatments and frequent comorbidities.
• Current FDA-approved medications for AD primarily manage symptoms, necessitating expert clinical judgment for optimal patient care.

Approach:

• This study proposes using reinforcement learning (RL) to model and learn optimal treatment strategies for AD patients from Electronic Health Records (EHR).
• A dataset of 1,736 patients from the Alzheimer's Disease Neuroimaging Initiative (ADNI) was analyzed, focusing on cohorts with hypertension and depression.
• The RL framework defined states (patient conditions), actions (medication choices), and rewards (MMSE scores) to simulate treatment decisions.

Key Points:

• The RL model demonstrated the ability to generate treatment policies that can potentially surpass clinician-driven regimens.
• Performance analysis showed that RL models achieved higher rewards with increased data sample sizes, indicating scalability.
• Specific RL algorithms like policy iteration and Q-learning were evaluated against clinician policies.

Conclusions:

• Reinforcement learning holds significant promise for developing AI-based decision support systems in managing Alzheimer's Disease.
• These systems can assist clinicians by providing optimized treatment recommendations based on longitudinal patient data and comorbidities.