Cost prediction of antipsychotic medication of psychiatric disorder using artificial neural network model

• Main Authors: Arash Mirabzadeh, Enayatollah Bakhshi, Mohamad Reza Khodae, Mohamad Reza Kooshesh, Bibi Riahi Mahabadi, Hossein Mirabzadeh, Akbar Biglarian
• Format: Article
• Language: English
• Published: Wolters Kluwer Medknow Publications 2013-01-01
• Series: Journal of Research in Medical Sciences
• Subjects: Linear regression; neural networks; psychiatric disorders; treatment cost
• Online Access: http://www.jmsjournal.net/article.asp?issn=1735-1995;year=2013;volume=18;issue=9;spage=782;epage=785;aulast=Mirabzadeh

Background: Antipsychotic monotherapy or polypharmacy (concurrent use of two or more antipsychotics) are used for treating patients with psychiatric disorders (PDs). Usually, antipsychotic monotherapy has a lower cost than polypharmacy. This study aimed to predict the cost of antipsychotic medications (AM) of psychiatric patients in Iran. Materials and Methods: For this purpose, 790 patients with PDs who were discharged between June and September 2010 were selected from Razi Psychiatric Hospital, Tehran, Iran. For cost prediction of AM of PD, neural network (NN) and multiple linear regression (MLR) models were used. Analysis of data was performed with R 2.15.1 software. Results: Mean ± standard deviation (SD) of the duration of hospitalization (days) in patients who were on monotherapy and polypharmacy was 31.19 ± 15.55 and 36.69 ± 15.93, respectively (P < 0.001). Mean and median costs of medication for monotherapy (n = 507) were $8.25 and $6.23 and for polypharmacy (n =192) were $13.30 and $9.48, respectively (P = 0.001). The important variables for cost prediction of AM were duration of hospitalization, type of treatment, and type of psychiatric ward in the MLR model, and duration of hospitalization, type of diagnosed disorder, type of treatment, age, Chlorpromazine dosage, and duration of disorder in the NN model. Conclusion: Our findings showed that the artificial NN (ANN) model can be used as a flexible model for cost prediction of AM.