Patient-specific prioritization of clinical guidelines is a promising innovation for complex patients, but the benefits and costs of prioritization have not been evaluated and prioritization programs are not yet available for widespread use. We propose a modeling project within the large cluster of diseases and treatments that interact in persons with type 2 diabetes mellitus. We will use a new kind of evidence-based computer simulation model, the Evidence-Based Medicine Integrator (EBMI). EBMI is designed to """"""""think"""""""" as an evidence-based clinician thinks, by combining the best user-specified trial evidence about treatment effectiveness with the best possible individualized estimates of patient risk. When fed a string of data from electronic patient records, EBMI identifies all available new treatments and dosage changes for each patient and simulates the long-term value of each of them. ? ? We will begin by adding adherence parameters to EBMI and writing a shell program that can simulate the effects of using EBMI to prioritize patients' treatments every quarter. These additions are necessary to realistically simulate the effects of computer-assisted clinical prioritization. ? ? We will then use EBMI to conduct an in silico (simulated) trial comparing (a) the use of a computer model to personalize and prioritize treatments versus (b) unprioritized adherence to written guidelines. We will simulate this comparison at two levels of clinical inertia, with high inertia set to approximate current behavior in the Kaiser Permanente medical care program (KPNW) and low inertia set at a level that might be induced by a powerful pay-for-performance scheme or by a strong institutional effort to maximize HEDIS scores. ? ? As the simulated trial is running, we will prepare and distribute the EBMI source code, an executable version of the model, and a de-identified version of the Kaiser Permanente (KP) dataset used to develop the model, using an open-source software development laboratory such as SourceForge.net. We will use a restrictive open-source license that, in addition to the usual requirements for continued access to the source code, requires users to provide a complete and exact run-settings file whenever they provide or publish modeling results. This will ensure that all assertions can be replicated (or disconfirmed) with ease by other workers. ? ? ? ?
| Blum, Shany; Vardi, Moshe; Brown, Jonathan B et al. (2010) Vitamin E reduces cardiovascular disease in individuals with diabetes mellitus and the haptoglobin 2-2 genotype. Pharmacogenomics 11:675-84 |
