UC San Diego

Machine learning and artificial intelligence (AI) hold the promise to innovate clinical practices and to improve quality of care. Naturally, the health research workflow may require modern adaptations to best capitalize the fast growth of AI. One challenge along the pipeline from health data to AI products is the balance between privacy-focused patients and access-focused researchers. It is thus crucial for scientists to overcome sharing barriers of health-research activities without doing harm to data security and privacy, so as to enable the meaningful use of electronic health records (EHR) in research. In particular, the sharing of research-oriented activities to benefit both patients and researchers can be facilitated through establishing a secured, blockchain-based informed consent infrastructure with which patients can grant data access to researchers. Other processes such as the management of clinical research activities in data consortia may also take advantage of such platforms. Following this feasibility of a robust, prolific data pipeline, the next step in health AI is the use of EHR for predictive analysis. Specifically, the COVID-19 pandemic has cast light on the critical value of predictive modeling in the fight against infectious diseases. A meaningful demonstration of EHR use in AI may include the construction of models to predict the risk of Clostridioides difficile infection. Last but not least, the efficiency and usability of cross-institutional research collaboration may also stand to benefit from workflow improvement, as it may enable better use of data from multiple sources. For instance, the credential-verification procedure to which researchers must abide to access external datasets may enhance efficiency through secured automation. Similarly, the use of privacy-preserving algorithms can be promoted by providing non-technical users with intuitive tools to participate in federated learning schemas. Together, it is seen that the health research workflow can be bolstered through fortifying the sharing framework of research-oriented activities, meaningfully utilizing EHR in predictive analysis, and improving the efficiency and usability of cross-institutional research. Such developments may boost scientific growth by keeping the right equilibrium among data quantity, data privacy, and research efficiency and usability, permitting the simultaneous expansion of patients’ autonomy and researchers’ innovation.