Perspectiveexperience Models of physiology populations and health care

Perspective/experience • Models of physiology, populations, and health care systems – Continuous variable, continuous time • Differential equations – Multiple diseases in single integrated model • E. g. cardiovascular disease, diabetes, cancers, congestive heart failure, etc – Written at clinical level • Physiological variables and pathways, signs and symptoms, patient behaviors, care processes, provider behaviors, logistics (e. g. , visits, admissions, tests, treatments, outcomes, utilization, costs) – Object oriented programming and grid computing • Enables “horizontal” and “vertical” expansion • Trying to create a “virtual world” • Validated by simulating epidemiological studies and clinical trials

Our experience • Used for “Policy” – Clinical trial design and prediction – Population-based policies • Guidelines, performance measurement, coverage policies, incentives (e. g. , P 4 P), priority setting, cost and cost-effectiveness analysis – Individual patient-physician decision making • Used by – – – – Voluntary health organizations (e. g. , ADA, ACS, AHA) Health plans (e. g. Kaiser Permanente, Humana) Insurers (e. g. Blue Cross Blue shield) National policymaking organizations (e. g. NCQA) Specialty societies and boards (e. g. internal medicine) Government agencies (e. g. CDC) Pharma and device manufacturers

Question 1 • How (policy) modeling has effected various research fields (success stories and mechanisms)? • Used by different types of organizations for different types of decisions – See previous slide • Several success stories – E. g. “A-L-L” program at Kaiser Permanente • Evaluation – Care of individual patients • Evaluation – Design of clinical trials – Drug portfolio analysis – Others

Question 2 • To what extent has the broader research community accepted modeling as a critical tool for driving research (what has worked and what hasn't)? • Use of modeling is still new and adoption is spotty – Initial reaction is skepticism – Mathematical modeling is not a “natural” part of decision making in medicine – Difficult for decision makers (and some other modelers) to understand the underlying mathematics • Success stories require considerable effort – Explain the model (e. g. , formulation, equations) – Validations • Success breeds success

Question 3 • In what ways can modeling further effect the broader research policy communities (how far can we go)? • Adoption depends on two main things – Credibility of the model • Clinical realism • Explanation of the model’s formulation and equations • Validation – Appreciation of the limitations of the alternative ways of making the decisions • Clinical trials usually not feasible • Clinical and administrative judgment very limited • As these issues are addressed, policy modeling will continue to grow

Question 4 • What are the major challenges to overcome (how do we get there)? • The two issues – Credibility of the model • Clinical realism • Understanding the mathematics • Validation – Appreciation of the limitations of the alternative ways of making the decisions • Clinical trials usually not feasible • Clinical and administrative judgment very limited • The availability of data (person-specific, longitudinal) • Standardization of validations – Validations non-mathematicians can trust