Home Medizin ChatGPT hilft bei der Entdeckung potenzieller Alzheimer-Behandlungen durch die Umnutzung von Medikamenten

ChatGPT hilft bei der Entdeckung potenzieller Alzheimer-Behandlungen durch die Umnutzung von Medikamenten

von NFI Redaktion

In a recent Proof-of-Concept study published in the journal npj Digital Medicine, researchers from the United States used Generative Artificial Intelligence (GAI) in the form of ChatGPT-4 to identify candidates for drug repurposing for Alzheimer’s disease (AD).

They found that GAI technologies can successfully integrate scientific online knowledge to help prioritize candidates for drug repurposing for disease treatment.

Study: Utilizing Generative AI to prioritize candidates for drug repurposing for Alzheimer’s disease with real clinical validation. Image Source: SuPatMaN/Shutterstock.com

Background

AD, a widespread and irreversible neurodegenerative disease, presents significant challenges to healthcare with limited treatment options.

Drug repurposing, exploring existing drugs for novel therapeutic applications, offers a faster and cost-effective alternative to developing new drugs for such health conditions.

This approach leverages established safety profiles, accelerates clinical implementation, and improves patient accessibility. Success depends on efficiently identifying promising candidates from a diverse pool of medications.

The conventional drug repurposing process requires comprehensive literature research, scouring various sources for evidence and synthesizing meaningful hypotheses from a vast search space. Optimizing this process is crucial to enhance its efficiency.

GAI demonstrates notable understanding and responsiveness, particularly in medical contexts like answering medical examination queries, clinical decision-making, and drug development.

Despite promising applications, its healthcare use requires rigorous scrutiny of its functional utility and reliability based on real clinical data due to concerns about potential information falsification.

The researchers of this study investigated the feasibility of utilizing GAI to identify drug repurposing candidates and clinically validated the recommended drug options using real data.

About the Study

The study conducted ten independent queries using ChatGPT-4 to identify drug repurposing candidates for AD. The queries were structured with prompts instructing the tool to list the top 20 medications based on their potential efficacy, excluding those originally developed for AD.

For clarity, the JavaScript Object Notation (JSON) format was specified, and a subsequent prompt was used to check and correct the list for distinguishability and correct ranking.

The queries aimed to promote differentiation between drugs developed for AD and other diseases, focusing on potential efficacy.

Clinical validation studies used data from electronic health records (EHR) from two datasets: the Vanderbilt University Medical Center (VUMC, n >3,000,000) and the National Institutes of Health All of Us Program (n = 235,000).

The data from the VUMC was anonymized. For each candidate drug, a retrospective cohort study began at the age of 65, excluding those with prior AD diagnosis, non-Alzheimer’s dementia, missing demographic data, or lacking EHR follow-up after age 65.

The AD diagnosis was based on specific ICD codes. Propensity Score (PS) matching (2:1) was employed considering gender, race, EHR length after age 65, and drug-specific comorbidities to create comparable drug-exposed and non-exposed cohorts.

The impact of multiple drug exposures on participants was not considered. The statistical analysis included Cox Proportional-Hazards regression models for survival analyses and a meta-analysis of hazard ratios (HR).

Results and Discussion

According to the study, the top three drug recommendations from ChatGPT-4 were Metformin (Antidiabetic), Losartan (Antihypertensive), and Minocycline (Antibiotic). Analysis of EHR data showed these drugs exhibited significantly lower AD risk after ten years.

While the results were limited by small sample sizes, Metformin showed treatment effects in a positive direction (HR <1). Additionally, Simvastatin and Pioglitazone showed potentially positive effects, though not statistically significant based on the VUMC and All of Us data analysis.

In the meta-analysis, Metformin showed a protective effect against Alzheimer’s disease (HR = 0.67), while Simvastatin and Losartan also displayed significant protective effects.

However, there was a contradictory alignment of effect estimates for Losartan between VUMC and All of Us datasets.

Inadequate case numbers complicated the evaluation of Bexarotene, Nilotinib, Minocycline, Candesartan, Rapamycin, and Lithium. Further investigations are needed to confirm these results.

ChatGPT-4 did not suggest any FDA-approved drugs for AD. The results demonstrate the tool’s effectiveness in drug repurposing based on its ability to follow instructions and quickly synthesize relevant information from the literature.

However, the results are limited by the tool’s reliance on medication prioritization frequency, potential issues with EHR data, limitations in statistical power, challenges in determining primary indications for some drugs, covariate imbalances, the inability to establish causality, and the need for continuous monitoring of the tool’s performance in drug repurposing.

Conclusion

In conclusion, the study demonstrates the potential and efficiency of ChatGPT-4 as an AI-based tool for drug repurposing, efficiently creating a promising drug list for testing in EHRs, with AD serving as a case study.

The results suggest that the tool can effectively retrieve and integrate information from various literature sources, providing an optimized framework that may be applicable to different diseases to uncover new therapeutic applications of existing drugs.

Related Posts

Adblock Detected

Please support us by disabling your AdBlocker extension from your browsers for our website.