This section describes literature review conducted in the field of MDSS in order to understand current state-of-the-art MDSS in the field of HIS. Throughout the process of literature review, it was observed there exists already huge amount of literature on how MDSS was built in the field of HIS. After carefully studying the entire literature, it was found that the existing MDSS were broadly classified into 3 systems.
• Forecasting binary outcomes
• Forecasting classification outcomes and
• Forecasting outbreaks.
In the next paragraphs, this thesis paper discussed various factors of 3 mentioned MDSS like main motivation for the implementation of new MDSS, different data mining (DM) algorithms used, techniques used to improve the…show more content… (Adeyemi, Demir, & Chaussalet, 2013); (Demir, 2014) argued that readmission rates should not be accounted for 30 days after surgery, moreover depending upon the disease type it varied from 30-90 days or other minor diseases had a short time frame of 14 days to longer time frame up to 1 year or up to 45-days from diagnosis.
To overcome this serious readmission rate problem, Center for Medicare and Medicaid services (CMS) imposed huge penalties on the hospitals, which had serious readmission rates (Amarasingham et al., 2010). Furthermore, most of the developed country governments devoted high importance to identify the patients who are at high risk of readmitting so that hospitals can plan the patient pathways. According to (Demir, 2014) they are:
a. Prevention and wellness promotion for relatively low risk patients
b. Supporting self-care interventions for moderate risk patients
c. Early intervention care management for patients with emergent risk and
d. Intensive case management for very high risk patients
Most of the researchers created many predictive models that predict the patient’s readmission with a simple “yes” or “no”. Few studies described in table 2, outline who conducted the study, data setting, data mining algorithm used and their accuracies.
Source Population setting Data Mining Algorithm ROC statistic
Anderson and Steinberg,1985 Medicare patients in the USA from 1974-1977 Logistic