Interpreting laboratory results is a demanding task in everyday medical practice. The project ‘Clinical decision support system for the interpretation of blood lab results and diagnoses based on them’, funded by the Founder Call, has now developed a clinical decision support system (CDSS) that assists doctors in making diagnoses.
The system is based on machine learning models trained with medical data from the University Hospital of Zurich from the last twelve years. It analyses age, gender, symptoms and blood laboratory values in two steps:
First, it evaluates 31 basic blood parameters and suggests diagnostic groups and further examinations.
In the second step, specialised models determine specific diagnoses.
The initial classification model achieves an accuracy of 90 per cent, while the specialised models range between 71 and 94 per cent. An automatic classification of blood cells has been added, enabling a better distinction between reactive and neoplastic diseases.
The project results are currently being prepared for publication in a scientific journal. The software is being further developed.