AI LOGIC & SYSTEM INTELLIGENCE
The current “state” of an Alzheimer’s patient is assessed by extensive testing: labs, cognitive tests, medical history, family history, genetic DNA analysis, medicines they are taking and supplements they are taking. ATCA’s practitioner’s get input from an extensive computer based analysis of all of this data. This enables the practitioner to create a set of personalized and prioritized interventions. Therefore when a genomic assessment shows a mutation indicating a strong likelihood for early-onset Alzheimer’s disease for a patient it will lead down a different treatment path than for a patient lacking that mutation. The data points from each patient are in the thousands. The data points already in our AI-Expert System are in the millions and each patient’s information is mapped against this treasure trove of information.
This broad set of assessment values serve as input to a rules engine. The rules engine is designed to 1) describe the condition of the person – their diagnosis and prognosis, and 2) recommend what should occur next. Recommendations include best-practice therapies for addressing the goals: prescriptions, non-prescription components, dietary changes, behavioral changes, even recommendations to get additional tests or more details of their medical history.
Not only are a person’s current test values used as input, but their entire history of inputs is used (as these can illustrate trends, ranges, effects, or correlations). The output of the engine serves as another input. In-puts from family members can also be used. Input across the entire population is used. The rules engine is complex, learns, has feedback loops, and creates output tailored to a person and their therapy team – and its output changes over time, as more input is received. This is a complexity far beyond the mental capabilities of a physician, thus requiring a software system as described here.
The system provides the functionality to seamlessly insert new information or components. For example, one can easily insert a new medication into the system as another potential recommendation or therapy. The new medication would include information that helps recommend or suggest the medication based on the data determined about the patient, such as side effects, treatment benefits, or conflicts with other medications or treatments. The system may also insert other new components or information, such as new therapies, treatments, recommendations, or medical tests. Incremental gains matter.
The system can determine appropriate mechanisms to target based on the data, and use analytics to identify therapeutics that have related performance characteristics that together form a combination that is highly effective relative to other treatments. This avoids less efficacious monotherapies and dynamically adjusts for success.
There is a significant body of training data to refine this system, that is, research data for retrospective study. One of these sources is being gathered as part of the worldwide Alzheimer’s disease Neuroimaging Initiative (ADNI). ADNI researchers collect, validate, and utilize data such as MRI and PET images, genetics, cognitive tests, cerebrospinal fluid (CSF) and blood biomarkers as predictors for the disease. Alzheimer’s AI-Expert System has access to this anonymized, license-free data, carefully collected over ten years from more than 3,000 human volunteers (http://adni.loni.usc.edu/).