Come back after 30 days!/ProjectOverview
 [|Home]
|
 [|Project Overview]
|
 [|Findings]
|
 [|Documentation]
|
 [|About Us]
|
At the moment of discharging from hospital,
Nurse: "Thank you for choosing our hospital, we hope that you have a speedy recovery!"
Patient: "Thank you, i feel much more at ease if i stay under the care of your hospital staff. If i feel any slightest discomfort, i will come back immediately ok!"
Nurse: *with an horrified face* "Oh.. uh.. actually, if its really something minor, you do not need to come back. But well, we can't stop any patients from visiting us... repeatedly"
Patient: "Yeah, so that's it. I will come back whenever i feel not at ease, even though it may be eventually minor. You know, just gotta be safe."
Hospital management often face such patients which constrains the operational capacity and efficiency. This leads to the motivation for studying this phenomenon.
Hospitals have been studying about the likelihood of patients readmitting within 30 days starting on the day of discharge primarily to reduce costs and operational overhead. It has grown to be a governmental concern as health systems, especially in UK, has decided to incentivize hospitals that abide by the rules and penalize those who did not manage to reduce their number of 30 days readmissions. Furthermore, for governments with welfare systems that provide for health care, patients readmitted within 30 days may be an avoidable expense if the hospitals were able to identify such patients during the first diagnosis. At the same time, various researchers claimed their superiority over other studies. Even though LACE (Length of stay, Acuity of admission, Comorbidity, Emergency department visits) index has been acknowledged as the gold standard in prediction of 30 day readmissions rates, authors claimed that their model perform better, often with caveats.
The outcomes of such a study allow hospitals to identify patients with a higher risk of readmissions and prescribe interventions in the form of house visits, or a wide-spectrum treatment in order to mitigate the problem. Data analytics, especially predictive ones, enable hospitals to do so. If effectively implemented, hospitals can reduce their costs and focus their limited resources to prevent avoidable readmissions. Consequently, hospitals may be able to stretch their resources to care for a larger number of patients, instead of serving readmitted patients.
To derive a model predicting the likelihood of a patient readmitting within 30 days using 2 approaches (i.e. random forest decision tree & multivariable logistic regression), along with its corresponding ROC curves to determine the predictive power of our models, thereby culminates in an evaluation of models.
