Figure 1. Data-driven simulation and forecasting in weather and oncology.
(A) Data assimilation systems from weather provide a model for long-term prediction of complex systems to enable predictive medicine. In weather, imaging and sensors capture high-throughput data profiling the atmosphere, which then calibrates mechanistic mathematical models that are based on the physical laws of the atmosphere. These mathematical models are used to predict future weather conditions, with increasing uncertainty as they are extended forward. Every 6-12 h, new data are assimilated with the forecasted state at that time to recalibrate the simulation and improve subsequent forecasts. Weather maps and model images from noaa.gov and satellite image of from GOES-16 weather satellite.
(B) Implementation of predictive medicine in neo-adjuvant platform clinical trials. Molecular and cellular profiling data from pre-treatment biopsies can be used to calibrate the states of cell-based mathematical models of tumors. These forecasts are then updated based on additional high-throughput data obtained from post-treatment surgical specimens to enable in silico clinical trials modeling the impact of new treatment strategies selected at later time points to overcome mechanisms of therapeutic resistance. Platform studies enable model calibration with multiple therapeutic agents. Cell-based models created with PhysiCell,2 and figure created with Biorender.com.