The research interests of John Kalantari, Ph.D., focus on novel applications of artificial intelligence (AI) to problems in biology and medicine, with a specific emphasis on the development of algorithms and computational frameworks for probabilistic Bayesian machine learning (ML), reinforcement learning, causal inference and complex adaptive systems.
The mission of Dr. Kalantari's research is to create AI and ML technologies that will transform medicine along two fronts — predictive analytics for real-time clinical decision-making and causal inference for preventive medicine. Among his current research efforts is the development of diagnostic and prognostic forecasting platforms that leverage experimental and clinical data for risk stratification, patient outcome prediction and individualized causal treatment effect estimation.
- Discovery of novel multi-omic therapeutic targets. Dr. Kalantari's laboratory focuses on identifying novel therapeutic targets using ML methodologies and multi-omics data.
- Development of precision medicine and causal inference methodologies. Dr. Kalantari and his colleagues are particularly interested in causal medicine: making the clinical decision-making process more data driven and evidence based. This requires developing probabilistic and causal inference methodologies for translating data into evidence-based, individualized treatment strategies that consider the causal effects of treatments on patient outcomes.
- Reverse engineering of the mechanisms of cancer progression. Dr. Kalantari's research group has invented the first reinforcement-learning algorithm for modeling the collective dynamics that define the disease mechanisms of tumor growth. These distinct disease mechanisms underlie the inter- or intratumor heterogeneity observed during disease progression and are essential to predicting tumor response or resistance to therapies.
Significance to patient care
Being able to provide accurate and accessible diagnoses and prognoses is a fundamental challenge in medicine. Recent advances in the fields of AI and ML have enabled diagnostic predictions, based on associations found in medical data, that identify diseases that are highly correlated with observed symptoms. However, to intervene as effectively as possible and to prevent further disease progression, health care professionals often seek to understand how and why a disease occurs so they can better explain a patient's symptoms.
To address these challenges, Dr. Kalantari and his colleagues are actively developing a clinical AI platform that uses causal reasoning to model and predict the most effective intervention for improving patient outcome. The ability to perform causal reasoning is essential for clinical decision-making. A clinical diagnosis is in effect a causal inference in which a health care professional reasons about causal models that relate putative causes to observed effects. Similarly, clinical prognoses represent a form of causal counterfactual modeling that uses "what-if" reasoning to predict how continuous-time trajectories (such as disease outcomes) will progress under different sequences of actions (such as health care interventions).