A New Normal: Digital Health in the Post-COVID Era
Stanford’s OTL closed out the 2021 fiscal year with 161 new licenses and 25 new startups, capping off a year of transition and a new economy that has emerged in the wake of the COVID-19 pandemic. Trends from these licensing deals point toward a growing industry interest in digital health, where new analysis and software tools have the potential to make healthcare more accessible, equitable, and effective.
A Digital revolution
Necessity is the mother of invention, and the COVID-19 pandemic has brought new urgency to longstanding needs for innovation in patient care. Even in the absence of physical distancing complications, accessibility has been a key challenge in both developing and delivering health care, reflected in the disparate health outcomes for underrepresented groups. According to the National Cancer Institute (NCI), an African American man is twice as likely to die from his disease as a white man following a diagnosis of prostate cancer, reflecting a combination of biological and socio-economic factors.
One key element toward equalizing health inequities is greater inclusion of racial minorities in clinical development- that is, more active recruitment of underrepresented patients in clinical trials for broader assessments of drug efficacy. A litany of historical and socio-economic factors limit the participation of minority groups in clinical studies- among them, the inconvenience of frequent travel to an often-distant clinic trial site.
Complex problems require complex solutions, and Stanford’s Nigam Shah is developing technology that could offer a solution for analyzing data from underrepresented groups. As a professor of medicine and biomedical data science, his laboratory has developed a time-aware search engine to mine electronic health records (EHRs), with the goal of using real world data to inform care for the poorly represented populations in clinical trials. The research group used the search engine to aide clinicians in decision-making at the bedside, combining the intelligence gathered via real-world EHRs with an expert physician consultation.
Shah is co-director of Stanford’s Center for Artificial Intelligence in Medicine & Imaging, which has a mandate to support the dissemination of AI technologies for the benefit of patients. Writing in the New England Journal of Medicine, Innovations in Care Delivery, Shah argues that using previously collected medical records to guide patient care has been a long-standing vision in precision health and one with the potential to transform medical practice.
“We cannot afford to miss the opportunity to use the patient data captured every day via EHR systems to close the evidence gap between available clinical guidelines and realities of clinical practice,” Shah says.
Working with OTL, the search engine has been licensed to Atropos Health, which launched under the leadership of Saurabh Gombar, an adjunct clinical assistant professor in the department of pathology and Chief Medical Officer for the company.
Support tools for clinical decision making are a fast-growing segment of the digital health revolution; other technologies available from Stanford include a clinical decision tool for allergic reactions to antibiotics from Clinical Associate Professor Anne Liu.
The New Normal
The benefits of an AI-enabled clinical studies aren’t limited to potentially excluded patients. Sponsors of clinical trials are tasked with the onerous responsibilities of patient recruitment and monitoring in pandemic conditions, which led to many studies grinding to a halt last year as hospital systems abandoned non-essential operations. Suspending a clinical trial over logistical issues represents a massive loss of time, data, and most importantly, a patient’s opportunity to receive an effective treatment.
The impact of these costs has not been lost on the financial sector; venture investment in clinical trial technologies more than doubled from 2019 to 2020, highlighting the role of COVID-19 in accelerating a new normal for trial operations. Opportunities for innovation abound in the space, as new ventures seek out methods for remote patient monitoring, optimized study design and enrollment, and heightened patient engagement- all from at least six feet apart. Non-profits have also taken notice of the opportunity for new technology to improve outcomes and equity, with UC Berkeley and UC San Francisco partnering last month to develop a joint program in computational precision health. Here at Stanford, the Center for Artificial Intelligence in Medicine and Imaging is developing a publicly accessible repository of AI-ready medical datasets in shared research hub, providing a resource to train AI-aided imaging models. The hub includes more than a million medical images with a huge diversity of data, which should enable researchers to unravel hidden biases that can easily slip into automated analyses.
Stanford professor of radiology Greg Zaharchuk has put medical images to work with the goal of enabling smaller clinical trials. He has developed a deep learning tool that can create virtual control arms for clinical studies, allowing study sponsors to recruit fewer patients. By analyzing existing medical images for conditions like stroke across a range of disease progression, the tool can map out a virtual course of disease progression. This allows patients to serve as their own control so fewer are enrolled in a study’s placebo arm, reducing a patient’s risk of being treated with placebo rather than a potentially beneficial therapeutic.
Adding these new bells and whistles to a clinical trial protocol serves to not only cuts costs and provide pandemic protection, but can also lower barriers around trial participation, as uncertainty over whether one might receive treatment or a placebo control is a contributor to clinical trial hestitancy. The ability to ensure remote participation in clinical trials is another key tool to future-proof against pandemic disturbances; technology from Chemical Engineering Professor Zhenan Bao helps solve this problem for cardiovascular studies by creating an ultrathin bandage to support the electrical leads used for electrocardiograms, ensuring that stable measurements are also comfortable for the patient with minimal disturbance to daily activities.
Looking forward
Improving clinical trial design is a feat in and of itself, but bringing digital health to the fore in drug design can offer even greater benefits. By starting clinical development with drug candidates that are pre-selected for personalized efficacy, fewer patients will be exposed to ineffective drugs and trials are more likely to succeed.
Integrating this strategy into the front end of drug development- by selectively enrolling patients that are likely responders- uses precision medicine to support smaller and less expensive clinical trials. This is a comparatively easy effort in genetic diseases, but complex disorders introduce additional heterogeneity that adds plenty of confusing ‘noise’ to a dataset.
Researchers at Stanford have nonetheless taken on the challenge, applying AI learning models to some of the most complex conditions in biology- neurological diseases. A group led by Psychiatry and Behavioral Sciences Professor Amit Etkin assessed electroencephalography (EEG) data from patients in a clinical trial to treat major depressive disorder, one of many psychiatric diseases marked by heterogenous and unpredictable responses to therapy. His group was able to develop a machine learning algorithm that could predict patient responses, effectively identifying a novel biomarker for the condition. Such training algorithms could have applications across neurological diseases, including post-traumatic stress disorder.
OTL licensed this technology to Alto Neuroscience, which emerged from stealth mode last month. If you are interested in exploring more Stanford technologies in digital health, check out our portfolios in clinical management and medical imaging!
