Constrained optimization for provider groups

A computer program product comprising a non-transitory computer readable storage medium having instructions encoded thereon that, when executed by a processor, cause the processor to:

• receive patient information describing a plurality of patients;
• receive health guidelines including requirements for filling certain health needs of patients or including a medical practice standard and describing one or more constraints to be satisfied for forming a provider group to treat health needs of the plurality of patients, the one or more constraints including a hard constraint that must be satisfied and a soft constraint that may be satisfied;
• identify a patient population based on the patient information and the health guidelines, the identified patient population having a plurality of health needs;
• receive health provider information describing a plurality of health providers;
• select a plurality of subsets of health providers from the plurality of health providers according to at least one of the hard constraint or the soft constraint, each provider of each subset capable of treating at least one of the health needs of the identified patient population, the hard constraint requiring a particular type of health provider be included in all subsets of the plurality of subsets, and the soft constraint requiring a particular number of health provider be included based on a size of a user population;
• input capabilities of each of the plurality of health providers into a first machine learning model;
• receive, as output from the first machine learning model, a respective weight for each respective health provider of the plurality of health providers;
• generate, for each subset of health providers of the plurality of subsets, a score using a second machine learning model trained based at least in part on the patient information, the health guidelines, and the health provider information, the score indicating a measurement of a capability of the subset of health providers for treating the plurality of health needs of the identified patient population, the score being determined at least in part based on the respective weights for each respective health provider in each respective subset of health providers;
• and
• determine a candidate subset of the plurality of subsets of health providers based at least in part on the scores.

A method comprising:

• receiving patient information describing a plurality of patients;
• receiving health guidelines including requirements for filling certain health needs of patients or including a medical practice standard and describing one or more constraints to be satisfied for forming a provider group to treat health needs of the plurality of patients, the one or more constraints including a hard constraint that must be satisfied and a soft constraint that may be satisfied;
• identifying a patient population based on the patient information and the health guidelines, the identified patient population having a plurality of health needs;
• receiving health provider information describing a plurality of health providers;
• selecting a plurality of subsets of health providers from the plurality of health providers according to at least one of the hard constraint or the soft constraint, each health provider of each subset capable of treating at least one health need, the hard constraint requiring a particular type of health provider be included in all subsets of the plurality of subsets, and the soft constraint requiring a particular number of health provider be included based on a size of a user population;
• inputting capabilities of each of the plurality of health providers into a first machine learning model;
• receiving, as output from the first machine learning model, a respective weight for each respective health provider of the plurality of health providers;
• generating, for each subset of health providers of the plurality of subsets, a score using a second machine learning model trained based at least in part on the patient information, the health guidelines, and the health provider information, the score indicating a measurement of a capability of the subset of health providers for treating the plurality of health needs of the identified patient population, the score being determined at least in part based on the respective weights for each respective health provider in each respective subset of health providers;
• and
• determining a candidate subset of the plurality of subsets of health providers based at least in part on the scores.