Detecting drug diversion in health-system data using machine learning and advanced analytics.

PURPOSE: The theft of drugs from healthcare facilities, also known as drug diversion, occurs frequently but is often undetected. This paper describes a research study to develop and test novel drug diversion detection methods. Improved diversion detection and reduction in diversion improves patient safety, limits harm to the person diverting, reduces the public health impact of substance use disorder, and mitigates significant liability risk to pharmacists and their organizations. METHODS: Ten acute care inpatient hospitals across 4 independent health systems extracted 2 datasets from various health information technology systems. Both datasets were consolidated, normalized, classified, and sampled to provide a harmonious dataset for analysis. Supervised machine learning methods were iteratively used on the initial sample dataset to train algorithms to classify medication movement transactions as involving a low or high risk of diversion. Thereafter, the resulting machine learning model classified the risk of diversion in a historical dataset capturing 8 to 24 months of history that included 27.9 million medication movement transactions by 19,037 nursing, 1,047 pharmacy, and 712 anesthesia clinicians and that included 22 known, blinded diversion cases to measure when the model would have detected the diversion compared to when the diversion was actually detected by existing methods. RESULTS: The machine learning model had 96.3% accuracy, 95.9% specificity, and 96.6% sensitivity in detecting transactions involving a high risk of diversion using the initial sample dataset. In subsequent testing using the much larger historical dataset, the analytics detected known diversion cases (n = 22) in blinded data faster than existing detection methods (a mean of 160 days and a median of 74 days faster; range, 7-579 days faster). CONCLUSION: The study showed that (1) consolidated datasets and (2) supervised machine learning can detect known diversion cases faster than existing detection methods. Users of the technology also noted improved investigation efficiency.

Replication-competent adenovirus-mediated suicide gene therapy with radiation in a preclinical model of pancreatic cancer.

In preparation for a Phase I trial, we evaluated the efficacy and toxicity of replication-competent adenovirus-mediated suicide gene therapy in combination with radiation in a preclinical model of pancreatic cancer. Human MiaPaCa-2 and PANC-1 pancreatic adenocarcinoma cells were found to be sensitive to the oncolytic effects of the Ad5-yCD/mutTK(SR39)rep-ADP adenovirus and also to the cytotoxic effects of the yeast cytosine deaminase (yCD) and herpes simplex virus thymidine kinase (HSV-1 TK(SR39)) genes in vitro. Combining Ad5-yCD/mutTK(SR39)rep-ADP-mediated suicide gene therapy with radiation significantly increased tumor control beyond that of either modality alone. Injection of Ad5-yCD/mutTK(SR39)rep-ADP in the dog pancreas at doses (10(12) virus particle (vp)) to be used in humans resulted in mild pancreatitis but not peritonitis or hepatotoxicity. Following administration of 9-(4-[(18)F]-fluoro-3-hydroxymethylbutyl)guanine ([(18)F]-FHBG), a positron-emitting substrate of HSV-1 TK, Ad5-yCD/mutTK(SR39)rep-ADP activity could be monitored non-invasively by positron emission tomography (PET). [(18)F]-FHBG uptake was readily detected in the pancreas but not in other major abdominal organs, indicating that little of the injected adenovirus disseminates to collateral tissues. These results demonstrate that Ad5-yCD/mutTK(SR39)rep-ADP-mediated suicide gene therapy has the potential to augment the effectiveness of pancreatic radiotherapy without resulting in excessive toxicity. Hence they provide the scientific basis for an ongoing Phase I trial in pancreatic cancer.