Detecting biomarkers associated with antipsychotic-induced extrapyramidal syndromes by using machine learning...

Area of Science:

• Pharmacology
• Clinical Chemistry
• Computational Biology

Background:

• Antipsychotic medications can cause extrapyramidal syndromes (EPS), leading to treatment cessation.
• Predictive biomarkers for EPS risk are currently lacking.
• Understanding EPS development is crucial for patient management.

Purpose of the Study:

• To identify laboratory parameters that predict the risk of developing antipsychotic-associated EPS.
• To leverage machine learning for biomarker discovery in EPS.
• To enhance patient safety during antipsychotic treatment.

Main Methods:

• Analysis of laboratory parameters from 390 hospitalizations of patients on antipsychotic medication.
• Application of machine learning, specifically random forests, to identify predictive biomarkers.
• Comparison of laboratory values between patients with and without EPS.

Main Results:

• Random forests model highlighted the importance of specific laboratory parameters in predicting EPS.
• Albumin (4.28%) and folate (4.09%) were identified as the most significant predictors.
• Lower mean levels of albumin and folate were observed in patients who developed EPS compared to those who did not.

Conclusions:

• Reduced levels of albumin and folate may be associated with an increased risk of developing EPS.
• These findings suggest potential biomarkers for predicting EPS risk.
• Further research is warranted to establish definitive cut-off levels for albumin and folate as predictive markers for EPS.