Article contributed by: Rich Christie, MD, PhD, Chief Medical Officer, AiCure
We’re continuing to see AI’s ability to transform clinical research and drug development, from its ability to streamline processes, to its capacity to drive efficiency. Even more remarkable is AI’s potential to propel precision medicine forward. AI can help the pharmaceutical industry achieve unimaginable levels of specificity and provide patients with tailored care to combat disease and improve their quality of life. Just as genomics revolutionized precision medicine in oncology over the last several decades, the behavioral measurements we’re now able to detect with AI tools have brought the industry to an inflection point in tailoring therapeutic interventions to individual patient needs. With the adoption of digital biomarkers and other AI- and predictive analytics-driven tools, we can detect specific, subtle behaviors associated with certain diseases with more precision than ever before.
In 2022 and beyond, these tools will continue to equip sponsors with a deeper understanding of disease progression and drug efficacy, and physicians with the insights they need to make informed interventions, help patients stay on track, and personalize treatment plans.
The role of novel, digital assessments
The challenge of precision medicine lies not only in its ability to match the right patient with the right therapeutic, but the ability to do so at the right point in their disease. Traditionally, clinical trials require in-person visits that only provide a restricted window into a patient’s condition. During these check-ins, physicians observe and note any changes in a patient’s emotional and physical condition, but these assessments fall short in capturing symptoms that may not be apparent at the time of the visit, or may be too subtle or infrequent to detect in real-time.
Video- and audio-based digital biomarkers can help achieve the level of timeliness and accuracy needed when capturing a patient’s response to treatment in ways that in-person assessments cannot. Looking at an individual’s disease trajectory, physicians can understand how medications are impacting a person’s quality of life through activities such as their ability to button their shirt or sign their name. They can also understand subtle visual and auditory symptoms, such as frequency of eye twitches or facial tremors. From there, it can be determined what intervention is going to provide the most value for that patient, and when it will be best to intervene. The ability to remotely identify these nuanced behaviors on reliable, compliant technology platforms can also augment physician decision-making regarding diagnosis and treatment.
The predictive power of AI in improving precision medicine
In addition to digital biomarkers, AI-powered predictive analytics can also help drive precision medicine by predicting patient behavior before trials even begin. By leveraging AI’s ability to predict a certain patient’s propensity to adhere, or not adhere, to a trial’s protocols based on their previous behavior, trial sites can focus on patients who may have the most trouble adhering to a treatment and tailor personalized resources towards them. This method can also help sponsors optimize patient pools and further inform precision medicine by understanding which patients will be able to contribute quality data for specific studies.
The challenges and promise of precision medicine
With the increased use of these AI-powered tools comes a renewed focus on the need for them to embrace inherent diversity in patient populations in a way that mitigates the impact of bias on patient care. If not managed properly, AI has the propensity to perpetuate unseen biases, and as an industry, we’ll need to be more alert and sensitive than ever to the impact of these biases on the applications of AI-driven systems. AI has the unique power to catapult precision medicine forward, helping sponsors better understand how medications work, their impact on specific patients, and bridge clinical findings into an increasingly diverse environment of real-world data collection. We must implement protocols that will help address these obstacles, continuously assessing AI models to ensure they’re generalizable and will work in the real world.
In 2022, as well, there will be ever-increasing amounts and sources of patient data. With that, precision medicine will become even more of a challenge as we try to translate these data into insights around patient experience, disease progression, and therapeutic impact. AI can not only help in filling these blind spots and serving patients the personalized care they need, but it can also advance our understanding of historically complex disease states and improve health outcomes of patient populations everywhere.
In order for AI to revolutionize drug development through its role in driving precision medicine, we’ll also need to continue using open-source data and algorithm platforms so that AI can be properly vetted, and the scientific community can communicate performance and opportunity around the use of these tools. For example, digital biomarkers are a unique resource for measuring patient behavior and improving precision medicine, but in many cases the proprietary nature of the algorithm involved inhibits broad adoption and collaboration around refining their clinical utility through validation. True transparency into the algorithms and lifting the veil on how they function is going to be key to their advancement and widespread adoption.