An Example of Application of Machine Learning Method to Laboratory Medicine: Determination of Newborn Reference Ranges

Abstract

Objective: Due to the difficulties in determining reference intervals with conventional methods, it determines them using modern machine learning methods. Methods: The results of the newborns' inorganic phosphorus, calcium, creatinine, neonatal bilirubin, and urea nitrogen tests, which were studied in the Dokuz Eylül University Central Laboratory for the years 2018-2019-2020, were obtained from the hospital database. The unsupervised machine learning algorithm we developed calculated test-specific age intervals and related reference intervals. Results: It was determined that the unsupervised machine learning method we developed is a new, contemporary alternative to indirect methods for determining reference intervals. With this method, the age ranges in which the test results showed significant variability were found in a high-resolution test-specific manner. Conclusion: Increases in computer processing power, new original artificial intelligence and machine learning-based algorithms, and databases that store large amounts of data offer a contemporary solution for determining reference intervals. In this study, an unsupervised machine learning algorithm solution based on mathematical and statistical foundations, which can determine age ranges, which is the basic step in the calculation of reference intervals, with high resolution is presented. By using the algorithmic method developed in the study, each laboratory will be able to calculate reference intervals suitable for their population and analytical methods in an easy, fast, safe, and economical way.

Keywords

• machine learning
• reference intervals
• newborn
• electronic patient records
• overlapping normal distributions