Application of unsupervised machine learning to identify areas of blood product wastage in transfusion medicine.

BACKGROUND: Wastage of blood products can be a significant cost to blood banks. However, the cause of wastage is often complex and makes it difficult to determine wastage-associated factors. Machine learning techniques may be useful tools to investigate these complex associations. We investigated whether unsupervised machine learning can identify patterns associated with wastage in our blood bank. MATERIALS AND METHODS: Data on red blood cells, platelets and frozen products were obtained from the laboratory information system of the Central Zone Blood Transfusion Services at Nova Scotia Health Authority. A total of 879 532 transactions were analysed by association rule mining, a type of machine learning algorithm. Associations with lift scores greater than 25 and with clinical relevance were flagged for further examination. RESULTS: Association rule mining returned a total of 3355 associations related to wastage. Several notable associations were identified. For example, certain wards were associated with wastage due to thawing unused frozen products. Other examples included association between smaller blood banks and evening work shifts with product wastage due to excess time outside the laboratory or returning products with high temperatures. CONCLUSION: This paper demonstrates the effective use of unsupervised machine learning for the purpose of investigating wastage in a large blood bank. The use of association rule mining was able to identify wastage factors, which can help guide quality improvement initiatives. This technique can be automated to provide rapid analysis of complex associations contributing to wastage and could be utilized in modern blood banks.

The successful implementation of an automated institution-wide assessment of hemoglobin and ABO typing to dynamically estimate red blood cell inventory requirements.

BACKGROUND: Blood bank inventories must balance adequate supply with minimal outdate rates. The day-to-day practice of ordering red blood cell (RBC) inventory usually involves manually comparing current inventory levels with predetermined thresholds calculated from historical usage and ordering the difference. To date, there have been no published methods for ordering RBC inventory based on laboratory characteristics of admitted patients. STUDY DESIGN AND METHODS: We designed and implemented a blood ordering algorithm to provide a more accurate measure of predicted RBC utilization in our institution. Cerner Command Language (Cerner Millennium) was used to extract and combine historical RBC unit usage, current inventory levels, and system-wide hematology values and blood groups. This report contains a suggested order based on current inventory, historical inventory data, ABO group, and the current "anemia index" for the institution. RESULTS: The mean daily total RBC inventory was significantly reduced after implementation (401.7 units vs. 309.0 units, p < 0.05). There was a significant reduction in monthly RBC outdates in this period (19.1 vs. 8.1, p < 0.05). The age of RBCs at time of transfusion was reduced as well. CONCLUSION: We developed a novel algorithm that automatically generates a suggested RBC inventory order using real-time hospital-wide survey of patient ABO typing, hematology values, and historical data. After implementation of the algorithm we demonstrated a significant reduction in daily inventory levels and RBC outdate rates.

Reflex and reflective testing strategies for early detection of pituitary dysfunction.

OBJECTIVE: The clinical presentation of pituitary dysfunction is typically variable and may often be insidious, resulting in delayed diagnosis by up to decades. The complexity of presentation and difficulty in pattern recognition of first line hormone tests result in challenges in early diagnosis of this condition. The aim of this study was to determine the impact of reflective testing and interpretive commenting on the early detection and management of such cases from primary care. METHODS: Prospective audit over 12 months in which first line pituitary target organ hormones were identified via a reflex algorithm in the laboratory information system. Selected tests were reviewed by a laboratory clinician and decision made on reflective testing and interpretive commenting based on available clinical information and previous result trends. Patients who had a laboratory intervention were followed up to determine the clinical outcome. RESULTS: Out of 1099 patients identified, additional testing was made for 214. Interpretative comments were subsequently added to reports of 196 patients, 48 (25%) of whom were referred to endocrinology and 35 (73%) of these were directly related to the laboratory intervention. Eleven other patients had outcomes related to the intervention. Pituitary related conditions (insufficiency and/or adenoma) were found in 29 patients, 24 of which were identified as a result of laboratory intervention. CONCLUSIONS: This study highlights the clinical value of laboratory intervention in aiding early detection of pituitary dysfunction and may avoid the disease burden of delayed management.