SBIR-STTR Award

Managed care organizations require primary care physicians to triage patients for skin cancer, yet many primary care physicians are unable to proficiently perform this function. Data suggest that primary care physicians correctly identify cancerous lesions less than 50% of the time, resulting in potentially disastrous health consequences for patients and increased health care costs. Triaging for skin cancer is a complex and difficult task, requiring primary care physicians to attend to important features of patients' skin lesions and history; sort these data; perform probability calculations; and follow complex lines of reasoning. Because computer programs can perform most of these functions faster and more reliably than physicians, such programs may help physicians triage for skin cancer. The goal of this project is to develop a prototype of a computer-based decision support tool to improve the accuracy of primary care physicians in triaging lesions suspicious for nonmelanoma skin cancer. During Phase I, we will develop a set of algorithms for triaging lesions suspicious for nonmelanoma skin cancer; develop a prototype of a computer-based decision support tool based on these algorithms; conduct a pilot test of the prototype with a sample of primary care physicians; analyze data; and write a final report. PROPOSED COMMERCIAL APPLICATION: This research is designed to produce a computer application that will be a commercially viable product. This computer-based decision support tool will improve patient health care while lowering health care costs. Potential purchasers of this product are managed health care organizations, primary care medical settings, and primary care physicians.

Thesaurus Terms:
computer assisted diagnosis, computer assisted medical decision making, computer assisted patient care, computer system design /evaluation, neoplasm /cancer diagnosis, skin neoplasm computer processing of clinical data, computer program /software, diagnosis quality /standard, early diagnosis, mathematics, primary care physician bioengineering /biomedical engineering, biomedical engineering, clinical research, human subject, statistics /biometryNational Cancer Institute (NCI)

Many primary care physicians are unable to proficiently triage skin lesions suspicious for cancer, compromising patients' health. Decision support software could help in the early detection of skin cancer, decreasing morbidity and mortality related to skin cancer misdiagnoses. It also could reduce unnecessary referrals to specialists, lowering health care costs. In Phase I, we demonstrated that decision support software can significantly improve physicians' ability to arrive at appropriate triage decisions for a variety skin lesions, including cancerous lesions. In Phase II, we propose to l) expand the decision support software to triage for pigmented lesions and a wider range of nonpigmented skin lesions; 2) incorporate case-based reasoning methods into the software to enable users to conduct a confidence check on their final triage decision and to help them make the correct decision at each branch point in the decision tree; 3) develop a teaching application based on the decision tree; 4) enable the software to produce for patients individualized information on skin cancer prevention and treatment; 5) interface the decision support software with the computerized patient medical record; and 6) test the effectiveness, utility, and feasibility of the decision support software with a sample of primary care physicians and patients. PROPOSED COMMERCIAL APPLICATIONS: This research is designed to produce comprehensive decision support software that will be a commercially viable product. Two large potential markets for this product are medical schools, which may use the software as a teaching aid, and medical settings, which may use the software as a clinical tool for practicing primary care physicians.