Part-5 Data Science Methodology From Deployment to Feedback

From Deployment to Feedback

Welcome to the data science methodology. Till now we have seen all 4 stages of data science methodology from Problem to approach, Requirement to collections, Understanding to preparation, Modeling to Evaluation. We have discuss amazing example with case study approach if you haven’t read this article series read from below links. and already read that go directly with this articles. In this article, You can learn about how to Model deploy and how to take feed back a model so model become more mature by the time.

Article Series :

1. Overview of Data Science Methodology
2. Part-1 Data Science Methodology- From Problem to Approach
3. Part-2 Data Science Methodology From Requirement to Collection
4. Part-3 Data Science Methodology From Understanding to Preparation
5. Part-4 Data Science Methodology From Modelling to Evaluation
6. Part-5 Data Science Methodology From Deployment to Feedback

#1) Deployment

• While a data science model provides an answer, the key to making the answer relevant to answering the initial question is to familiarize people with the product tool. In a business scenario, stakeholders have different characteristics, such as: The solution owner, marketing, application developers and IT administration.
• Once the model has been evaluated and the Data scientist is convinced that it will work, it will be used and subjected to the final test.
• Depending on the purpose of the model, it can be extended to a limited group of users or in a test environment to increase confidence in the application of the result to global use.

Case Study:

• Let’s take a look at the case study of the implementation application “In preparation, To provide the solution, the next step was to gather the knowledge of the stakeholder group responsible for the design and management of the intervention program to reduce the risk of readmission.
• In this scenario, entrepreneurs have translated the results of the model so that clinical staff can understand how to identify high-risk patients and design appropriate interventions.
• Of course, the objective was to reduce the risk of readmission of these patients within 30 days of discharge. During the operational requirements phase, the intervention program director and her team looked for an application that could assess the risk of heart failure almost automatically in real time.

• It should also be easy for clinical staff to use, preferably through a browser and tablet-based application that any employee could carry with them. These patient data were generated throughout the hospital stay. It will be generated automatically in a format required by the model and each patient will be noticed shortly before discharge.
• Then, doctors would have the most up-to-date risk assessment for each patient to help them choose which patients to treat after discharge.
• As part of providing the solution, the intervention team would develop and offer training to clinical staff.

• In addition, in collaboration with IT developers and database administrators, monitoring and monitoring processes should be developed for patients receiving the intervention, so that the results can go through the feedback phase and the model can be mature over time.

• This Map is an example of a solution implemented through a Cognos application( IBM Cognos Business Intelligence is a web-based integrated business intelligence suite by IBM). In this case, the case study focused on the risk of hospitalization of patients with juvenile diabetes. Similar to congestive heart failure, he used the classification of the decision tree to create a risk model that would form the basis of this application.

• The map provides an overview of hospital risk nationwide, with a planned interactive risk assessment for different patient conditions and other characteristics. This above image provides an interactive summary report on the risk per patient population in a given node of the model so that doctors can understand the combination of conditions for that subset of patients.

• This report provides a detailed summary of a single patient, including details of the patient’s history and expected risk, and provides the doctor with a brief summary.

#2) Feedback

• Feedback ! After playing, user feedback help refine the model and evaluate its performance and impact. The value of the model depends on the successful integration of feedback and customization whenever the solution is needed.
• Throughout the methodology of data science, each step paves the way for the next. By making the methodology cyclical, you ensure a refinement at each stage of the game.
• The feedback process is based on the idea that the more you know, the more you want to know.

• Once the model has been evaluated and the data scientist trusts that it will work, it will be implemented and will undergo the final test: its real use in real time in the field.

Case Study:

• Now let’s review our case study to see how the part of the feedback methodology is applied.
• The feedback phase plan included the following steps: First, the review process would be defined and established, with the overall responsibility of measuring the results of a flight risk model of the heart failure risk population.Clinical management has overall responsibility for the review process.
• Second, patients with heart failure who receive an intervention would be monitored and their readmission results recorded.
• Third, the intervention would be measured to determine its effectiveness in reducing the number of readmissions.
• For ethical reasons, patients with heart failure would not be divided into controlled groups and treatment groups. Readmission rates are compared before and after the implementation of the model to measure the impact.

• After deployment and feedback, the impact of the intervention program on readmission rates will be reviewed after the first year of implementation.
• Then, the model would be refined based on all data compiled after the implementation of the model and the knowledge acquired in these steps. Other improvements include the inclusion of information on participation in the intervention program and possibly the refinement of the detailed pharmaceutical data model.
• If you remember, data collection was initially delayed because drug data was not available at that time. However, after feedback and practical experience of the model, it can be said that adding this data can be worth the investment of time and money.We must also consider the possibility of new adjustments during the feedback phase.

• In addition, response actions and processes are reviewed and probably refined according to the experience and knowledge acquired during the initial implementation and feedback.
• Finally, the refined model and intervention would be redeployed, and the feedback process would continue throughout the intervention program.

Conclusion :

I hope you are enjoying this article series Thanks for reading…!!! Happy Learning…!!!

References :

1. https://www.coursera.org/learn/data-science-methodology