Prediction of Pulmonary Function Test Values in COPD Patients Using Artificial Intelligence Architecture- Clinical Decision Support System

Objective: To develop a model that can predict pulmonary function test (PFT) values in individuals with chronic obstructive pulmonary disease (COPD) using an artificial neural network (ANN). Method: Levenberg-Marquardt algorithm was used. For performance testing, the ANN was trained using the Mean Sequential Error (MSE) method. While age, sex etc. of the individual were input data, PFT value was output data. The data required to test this model were 29 patients diagnosed with COPD, aged between 40 and 70 years, who were referred to Malatya Training and Research Hospital Chest Diseases Polyclinic. A triple cross validation test was used to calculate the performance of the system. The performance parameter was determined using the accuracy parameter. Results: A triple cross validation test was used to calculate performance of system. Accuracy parameter was used as performance parameter. In designed model, average success rates were determined for each PFT value and total average success rate was evaluated as 97.40%. Conclusion: With this system PFT values can be easily determined. It is believed that the system will help in the management of dyspnoea, planning, creating an exercise treatment programme and maintaining quality of life. Keywords: Artificial neural network, Clinical decision support system, COPD.