Corey Lester - US grants

PROJECT SUMMARY Medical errors are the 3rd leading cause of death in the United States behind cancer and cardiovascular disease. The largest proportion of medical errors involve medications. Medication errors result in 3 million outpatient medical appointments, 1 million emergency department visits, and 125,000 hospital admissions each year. Astoundingly, over 4 billion prescriptions are dispensed every year in the United States alone. Although dispensing error rates are generally low at 0.06%, the sheer volume of dispensed medications translates to 2.4 million incorrectly dispensed medications each year. In the pharmacy, dispensing errors arise when pharmacists do not detect that the medication filled inside a prescription vial is different from the medication ordered on the prescription's label. These dispensing errors can result in patient harm, added strain on the healthcare system, and costly legal action against the pharmacy. Machine intelligence (MI) can be employed to assist in the verification process to help avoid dangerous and costly pharmacy dispensing errors.4?6 However for the human-MI partnership to function optimally, the MI should be capable of conveying accurate information that encourages providers to make sound cognitive decisions such that optimal trust is maintained, and temporal and cognitive demand is reduced. Imperative to this goal is to design MI from which interpretable information can be extracted, convey this information in an effective manner and calibrate user's trust in MI as either over-trust or under-trust can lead to near miss and incident errors. This proposed project will further our knowledge for designing interpretable MI outputs and inform the development of MI models that encourage pharmacy staff to make sound clinical decisions that lead to better patient outcomes while improving work-life at lower costs of care. This study develops interpretable MI methods in the context of medication images classification and designs effective MI advice and reasoning that lead to lower cognitive demand and increased trust in the MI. Our hypothesis is that interpretable MI will lead to improved work performance and more calibrated trust compared to uninterpretable M. The objectives of this proposal are to: 1) design interpretable machine intelligence to double-check dispensed medication images in real-time; 2) evaluate changes in pharmacy staff trust due to the long-term use of interpretable machine intelligence; and 3) determine the effect of interpretable machine intelligence on long-term pharmacy staff work performance.