Algorithm-based laboratory test ordering and interpretation of high-volume laboratory tests: a data-driven approach

Laboratory test ordering is one of the most important diagnostic procedures in medicine and there is a continuous increase in the number of laboratory tests ordered by physicians. Besides the increasing strain that the growing number of laboratory tests poses on healthcare budgets, there is concern about the rates of inappropriate test ordering. Clinical decision support (CDS) systems have shown to improve appropriateness and decrease volume without increasing diagnostic error. Most CDS systems used in laboratory test ordering are expert systems using algorithms based on guidelines or recommendations. The aim of this project is to complement current CDS systems with knowledge derived from big data to assist physicians in laboratory test ordering. We will use available data sets to analyze the results and clinical consequences for five high volume laboratory tests across health care levels. We will develop algorithms focusing on common indications or conditions to guide laboratory test ordering for these 5 tests. These algorithms will then be tested within a CDS system. This project will result in a CDS system containing a rule-based expert system and additional machine-learning based algorithms which can be implemented on an national level.

Keywords:decision support, artificial intelligence, laboratory test ordering

Disciplines:Decision support and group support systems, General diagnostics, Machine learning and decision making