COMPARING CLINICALLY BASED VERSUS ALGORITHM-GENERATED INTERVENTIONS IN THE CONTEXT OF ALZHEIMER’S DISEASE (AD) PREVENTION

Abstract

The Alzheimer's Prevention Clinic (APC) at Weill Cornell Medicine utilizes precision medicine interventions to reduce individual risk towards developing AD. The APC provides direct clinical care to people who undergo a comprehensive assessment and receive evidence-based, individualized interventions applying principles of pharmacogenomics and nutrigenomics, with the goal of stabilizing or improving cognition and blood biomarkers. Muses Labs follows a similar paradigm in the absence of an in-person clinical visit, using an informatics platform that receives and processes multi-sourced data to provide a targeted intervention plan known as the AD1 Protocol. In this pilot study, five subjects (ages 51–65) from the APC prospective cohort (n=377) were recruited to submit their clinical data to Muses Labs. Recommendations given by the APC neurologist were compared to artificial-intelligence generated recommendations from AD1. For each APC subject, an average of 334 data points (e.g., clinical history, labs, biometrics, cognition) were assessed compared with 1,029 data points assessed by Muses Labs for interpretation (sourced from genome, medical history – including comorbidities, medications, allergies, and vaccines – labs, and cognitive testing). The APC analyzed 58 blood biomarkers and 4 genetic risk factors, while Muses Labs processed an average of 107 blood biomarkers and ∼1,000 single nucleotide polymorphisms (SNPs). APC recommendations averaged 3.8 supplements/vitamins, 13.6 dietary modifications, and 8.4 lifestyle changes per subject, while AD1 averaged 16.8 supplements/vitamins, 5.2 potential supplements/vitamins, 9.2 dietary modifications, 5.8 lifestyle changes, and 4.4 additional diagnostic tests. AD1 reports averaged 7,504.2 words per person, while APC interventions averaged 1,272.8 words, both excluding laboratory results. Given the complexity of emerging precision medicine interventions for AD prevention, novel and collaborative efforts using both clinical informatics and physician-based approaches may offer a more comprehensive set of potential therapeutic recommendations for those seeking to reduce risk. Using consistent software-based decision trees that utilize emerging evidence could facilitate the application of precision medicine in the clinic, expand the number of potential interventions, and make complex protocols readily accessible to a large number of physicians and patients worldwide. Further study is warranted to evaluate the longitudinal effectiveness of these augmented preventative approaches.