By Greg Licholai | Forbes.com | Dec 2, 2019, 10:00am
Imagine having to choose from over 14,000 different treatment scenarios to decide which drugs might be best for a child or a loved one affected by epilepsy. This is what faces many families according to the experts at Stanford and doc.ai who have announced a new type of clinical trial using artificial intelligence (AI). The project’s goal is to help make the process more scientific using population data and less prone to lengthy individual trial-and-error. Researchers are analyzing medications, side effects, genomic information, environmental exposures, activity and even physical traits. This type of work produces vast amounts of information and requires so much processing power that it can only be performed by the latest AI systems.
The trial is being sponsored by Silicon Valley health-tech start-up doc.ai, which already has launched AI-based studies for Inflammatory Bowel Disease and Allergies. The novel approach has led to major institutional attention and participation. The company has teamed with Anthem who describes their collaboration as a “virtuous cycle” that lets the insurer benefit from A.I. technologies and help lead to more empowered patients.
The novel approach might help solve the dilemma of comparing multiple drugs at once. The principal investigator of the AI epilepsy trial is Dr. Robert Fisher, MD, PHD, a respected professor of neurology and neurological sciences at the Stanford University School of Medicine and director of the Stanford Epilepsy Center. Dr. Fisher says the study study is not designed to recommend specific drugs, only see if the method to do so is accurate. Factorial studies have been historically used to perform direct head-to-head comparison of medications. For example, testing the effects of two different medications requires 4 separate groups or “study arms.” Since epilepsy has so many drug combinations possible it would be unrealistic to compare 25 drugs using traditional a traditional design.
The new trial’s goal is to help determine the precision of epilepsy treatment options incorporating many “real world” variables. It uses the predictive capabilities of AI to figure out the most effective choices given the complexity of the disease and relatively high number of treatments.
At the end of the study patients will be given a report including charts and tables, seizure diaries, and summaries of past medication histories to help provide a full picture of their epilepsy. Together with doctors, they will be able to review both population results as well as individual effects.
Epilepsy is a disease of recurring unprovoked seizures and can be a devastating, life-altering condition. More than 65 million people are effected around the world. In its severe forms, it can shorten lifespan, cause bodily injury, neuropsychological and psychiatric impairment, not to mention the significant social disability. It spares no race, gender or ethnicity. To add to the complexity there are two major types of epilepsy called focal and general, with over a dozen subtypes that affect adults and children.
There are over 25 anti-epileptic drugs approved for the many different varieties of seizure disorders. The medications have different levels of tolerability, side effect profiles and mechanisms of action. One of the major treatment difficulties is knowing which medication would work best for any particular patient. Guidelines recommend starting on a single drug as monotherapy, however it is common practice to use multiple drugs as some patients have continue to have symptoms.
Experts assert that physicians should use evidence-based decision making when recommending treatment for epilepsy. Expert panels are often brought together to issue guidelines on the ever-changing standards of care. However specialists lament that it is “impossible to develop an evidence-based guideline aimed at identifying the overall optimal recommend” therapy due to “absence of rigorous comprehensive adverse effects data” from standard clinical trials.
The doc.ai Stanford trial shows how healthcare is being advanced by innovations such as linking large data sets and using AI to perform predictive analytics.