Operational Data Classification Modeling in Weka -- 2
Budget: $10 – $30 USD
The database was collected by a health-care organization and contains data on patients’ visits to doctors, lab tests done, diagnoses, etc.
Presently, the current practice is to consider a patient as being at high risk of becoming diabetic if this patient ever had a fasting blood glucose (FBC) level of above 110.
Your task is to develop models that would be used by the health care organization for different purposes.
1. Your team’s first goal is to allow the health-care organization select different subset of patients for their coach-based prevention programs based on the patients likelihood of developing diabetes within a 3-year window. At present, the healthcare provider is selecting patients for prevention program based on the abovementioned existing criterion for risk. Your aim is to explore the possibility of identifying more effective groups for intervention -- groups of patients who are at a higher risk than those selected presently. Presently, state-of-the-art prevention programs can reduce the chance one develops diabetes by 60% by changing the patients lifestyle, including physical activity and diet.
To assess ay cost savings you may achieve, you note the following:
• The organization’s prevention program is effective as reported (60%
reduction in patients developing diabetes).
• The annual medical expenditures per patient directly attributed to
diabetes $7,900
• Per-patients annual prevention program cost is $600
Using the information above, design a solution for selecting patients for intervention program such that the total costs savings are maximized. Discuss how the strategy (and costs) change as the cost of intervention/prevention varies. Also compare the savings of your proposed solution, to the cost savings that can be achieved if patients are selected for the intervention program using the current threshold criterion of Fasting Blood Glucose (FBC ) above 110.
2. Consider the objective of simply predicting who would become diabetic, namely simply classify patients as either the “diabetic” or “non-diabetic”. Discuss what is the most common type of error your model makes, and perhaps why do you think this error is common.
3. Discuss what are the most distinctive set of characteristics of patients who develop Type II diabetic as compared to patients who do not.
For all the tasks above consider alternative approaches to achieve the best results and compare with the current best practice.
Deliverables:
Your report should include a clear description of the problem, a detailed explanation of the approach(es) you have considered to address it, the rationale for each approach, and your recommendations. Please discuss any alternatives you considered in the process (such as alternative modeling techniques, different sets of variables, etc.), and demonstrate the benefits of the alternatives selected.
Your report should include an evaluation of the expected benefits of your approach on health care expenditures and demonstrate whether it is superior to the selection of patients for intervention based on the existing FBC criterion.
Consider also including any interesting insights you have gained through your work.
please use tool "weka"
Presently, the current practice is to consider a patient as being at high risk of becoming diabetic if this patient ever had a fasting blood glucose (FBC) level of above 110.
Your task is to develop models that would be used by the health care organization for different purposes.
1. Your team’s first goal is to allow the health-care organization select different subset of patients for their coach-based prevention programs based on the patients likelihood of developing diabetes within a 3-year window. At present, the healthcare provider is selecting patients for prevention program based on the abovementioned existing criterion for risk. Your aim is to explore the possibility of identifying more effective groups for intervention -- groups of patients who are at a higher risk than those selected presently. Presently, state-of-the-art prevention programs can reduce the chance one develops diabetes by 60% by changing the patients lifestyle, including physical activity and diet.
To assess ay cost savings you may achieve, you note the following:
• The organization’s prevention program is effective as reported (60%
reduction in patients developing diabetes).
• The annual medical expenditures per patient directly attributed to
diabetes $7,900
• Per-patients annual prevention program cost is $600
Using the information above, design a solution for selecting patients for intervention program such that the total costs savings are maximized. Discuss how the strategy (and costs) change as the cost of intervention/prevention varies. Also compare the savings of your proposed solution, to the cost savings that can be achieved if patients are selected for the intervention program using the current threshold criterion of Fasting Blood Glucose (FBC ) above 110.
2. Consider the objective of simply predicting who would become diabetic, namely simply classify patients as either the “diabetic” or “non-diabetic”. Discuss what is the most common type of error your model makes, and perhaps why do you think this error is common.
3. Discuss what are the most distinctive set of characteristics of patients who develop Type II diabetic as compared to patients who do not.
For all the tasks above consider alternative approaches to achieve the best results and compare with the current best practice.
Deliverables:
Your report should include a clear description of the problem, a detailed explanation of the approach(es) you have considered to address it, the rationale for each approach, and your recommendations. Please discuss any alternatives you considered in the process (such as alternative modeling techniques, different sets of variables, etc.), and demonstrate the benefits of the alternatives selected.
Your report should include an evaluation of the expected benefits of your approach on health care expenditures and demonstrate whether it is superior to the selection of patients for intervention based on the existing FBC criterion.
Consider also including any interesting insights you have gained through your work.
please use tool "weka"