How can predictive models impact business decisions?
Otherwise, one is forced to look at millions of rows every day and make decision. Human brain has non linear thinking to do this but, memory is limited. so we will end up doing the same thing over and over again. Predictive algorithms free human brain from its limitations and performs algorithms on computers do what human brains are not capable of- looking millions of records for each of at least three years and find solutions such as who will buy or which doc is likely to perform unnecessary surgeries. This is what one of the respected reviewer has written for "analytics at work" book.
At least in Healthcare, insurance companies know which hospital …show more content…
This knowledge is used to manage patient condition, for example, someone with fasting sugar of 115 is borderline diabetic patient. The patient are advised not just do exercise and stop eating rich food, but also "what is Statins" and given the genetic background, to what extent behavioral modifications will work.
Hence Predictive Analytics makes care that is patient centric. Patients are as much or even more of the treatment equation as Providers and Hospitals. Predictive Analytics can easily reveal which patients are likely candidate for re-admissions. Then providers make sure that patients and caregivers understand what the healthcare needs will be when they leave hospital.
Gone are the days that patients were discharged with prescriptions and drugs to go with. Now, before discharge, patients are asked if they are understand what medicines and why they are taking it? In other words Predictive Analytics also means communication with patients for bottom line improvement. After all, there is 2% penalty from each patient's revenue if that patient is readmitted more than 3 times in a month/ 30 …show more content…
is it the fault of data? To a little boy with hammer, the whole world looks like a nail... So a good Algorithm is subjected to database of Encyclopedia Britannica then the result comes out as "everyone before 1900 was famous!". One of the basic requirements of Predictive Analytics is not just volumes of data, but also understanding of biases in data. After understanding the biases, try to find the repercussions of false positives and negatives before touching the data. Meaning simulating a few patient conditions by ignoring at least one ICD 9 disease conditions. This means creating biases and getting better views from data; lack of data understanding is what fails Predictive Analytics. Given the need of data coming from as many different sources as possible to represent the use case in best possible way quantitatively, efforts in Data Understanding need to become