JAKCalc: A machine-learning approach to rationalized JAK2 testing in patients with elevated hemoglobin levels

Area of Science:

• Genetics
• Bioinformatics
• Hematology

Background:

• Janus Kinase-2 (JAK2) testing for erythrocytosis often yields low positive results, indicating a need for improved diagnostic strategies.
• Current testing methods may not be sufficiently discerning, leading to potential overutilization and increased healthcare costs.

Purpose of the Study:

• To introduce an artificial intelligence (AI) application for a more rational approach to JAK2 mutation testing in erythrocytosis.
• To evaluate the performance of machine learning models in identifying JAK2V617F mutations.

Main Methods:

• Utilized genetic testing data from 458 erythrocytosis cases (2017-2023) meeting WHO criteria for JAK2V617F mutation testing.
• Trained and tested various machine learning models, including Random Forest (RF) and Gradient Boosting (GB), using Python.
• Analyzed complete blood count parameters to identify differences between mutation carriers and non-carriers.

Main Results:

• JAK2V617F mutation was identified in 13.3% of the cases.
• The Random Forest (RF) model achieved 100% precision, recall, F1-score, accuracy, and area under the ROC curve.
• Gradient Boosting (GB) also demonstrated high performance, outperforming existing algorithms.

Conclusions:

• AI-driven machine learning models, particularly RF and GB, offer a superior and accurate method for identifying and classifying JAK2 mutations in erythrocytosis.
• These advanced models have the potential to significantly reduce unnecessary JAK2 testing and associated healthcare expenditures.