Probo: Smart Approval System

Objective: The results of 60% of the tests performed in biochemistry labs were evaluated using machine learning to provide an additional expert opinion on the approval status.

Method: A two-pronged approach has been developed to ensure a comprehensive evaluation of patients’ health status. This approach involves the use of automatic and smart approval systems. The automatic approval system is rule-based and is mainly deployed to identify any malfunctions caused by devices. Once the automatic approval evaluation is complete, the approval status is evaluated a second time using Support Vector Machines (SVM), Multi-layer Perceptron (MLP), and Random Forest (RF) – traditional and convolutional machine learning methods – to determine the instantaneous status of the patient. The experiments conducted using these systems have shown a success rate of 99% across 27 tests.

Innovative Features

• In the automated approval system, test results are scanned alongside patient information. Results with low compatibility trigger further action, such as retesting or referral to a specialist.
• Reducing error rate
• Increasing the speed of work in the laboratory
• Eliminating differences in decision-making among experts ensures that standard decisions are made consistently.
• Reducing the time spent on test results that do not require expertise and enabling the focus on test results that require domain knowledge and expertise.
• LIMS integration