Post discharge risk prediction

A system for predicting post-discharge risk, the system comprising:

• at least one processor configured to:
• access frames of video captured during a specific surgical procedure on a patient;
• access stored historical data identifying intraoperative events and associated post-discharge outcomes, the stored historical data including a hypergraph, wherein a hyperedge of the hypergraph connects a plurality of the intraoperative events with a respective outcome of the associated post-discharge outcomes and indicates a particular probability of the respective outcome occurring after discharge following surgical procedures that included the plurality of the intraoperative events;
• analyze the accessed frames to identify, using a machine learning model and based on information obtained from the historical data, at least one specific intraoperative event represented in the accessed frames, the machine learning model being trained using example training data;
• determine, based on the information obtained from the hypergraph and the identified at least one specific intraoperative event, a predicted outcome after a discharge of the patient following the specific surgical procedure;
• and
• output the predicted outcome in a manner associating the predicted outcome with the patient.

A non-transitory computer readable medium containing instructions that, when executed by at least one processor, cause the at least one processor to execute operations enabling prediction of post-discharge risk, the operations comprising:

• accessing frames of video captured during a specific surgical procedure on a patient;
• accessing stored historical data identifying intraoperative events and associated post-discharge outcomes, the stored historical data including a hypergraph, wherein a hyperedge of the hypergraph connects a plurality of the intraoperative events with a respective outcome of the associated post-discharge outcomes and indicates a particular probability of the respective outcome occurring after discharge following surgical procedures that included the plurality of the intraoperative events;
• analyzing the accessed frames to identify, using a machine learning model and based on information obtained from the historical data, at least one specific intraoperative event represented in the accessed frames, the machine learning model being trained using example training data;
• determining, based on the information obtained from the hypergraph and the identified at least one specific intraoperative event, a predicted outcome after a discharge of the patient following the specific surgical procedure;
• and
• outputting the predicted outcome in a manner associating the predicted outcome with the patient.