![]() | Only 14 pages are availabe for public view |
Abstract Since the wide use of computer aided systems in healthcare sector, vast amount of medical databases established. Those historical data used to set up descriptive and predictive applications using data mining and intelligent techniques. As a result, those techniques successfully built models which help administration and decision makers to enhance hospitals performance management in (treatment cost control, staff allocation and evaluation, demographic and market trends, quality assurance, process efficiency and improving healthcare quality). As well as, clinical decisions (disease diagnosis, medication plan, diseases early detection, mortality early detection, detecting high risk patients, etc.) According to the World Health Organization (WHO), one of the performance measurements and monitoring indicators is the hospitals’ length of staying and it also used to evaluate both financial and medical performance Prolonged Length of Stay (LOS) in Intensive Care Unit (ICU) leads to consuming hospitals resources as manpower and equipment. Moreover increase patients’ recovery duration and raising the probability of death during accommodation or after discharging. So, predicting LOS aims to best resources utilization and medical team allocation. Additionally, helping the healthcare specialists for more effective medical decision making. Increasing cost for ICU bed equipment and operating expenses lead to shortage in ICU beds. Therefore, providing health care providers with information about expected discharge dates helps more patients waiting for an empty bed as well as accounting department can recognize the amount of the initial insurance paid in cash for patients who do not have medical insurance. Furthermore, healthcare insurance companies can evaluate the expected cost for their clients and the quality of healthcare they receive. Researches provided in this area are interested in predicting the LOS for a particular diagnosis or complains for the patients who suffer from cancer, diabetes or even after a specific surgery; therefore, researches could afford having accepted prediction accuracy, especially when specific features in the evaluation and modeling were in use. |