Team Name
TEAM INGENUITY
Timeline
Spring 2020 – Summer 2020
Students
- Devi Tripathy
- Chelsea May
- Wei Shi
- Sudeep Bhadel
- Avijeet Adhikari
- Bijay Raj Raut
- Bhupendra Ramdam
- Amit Thing
- Edwin E Popaja
Sponsor
State Farm
Abstract
Currently, how life insurance rates are determined is not done with a very holistic view of the person’s overall health at the present time. Life insurance agents look at many aspects of the customer’s health, but not how healthy or unhealthy their day to day lifestyle is. Does the customer exercise? If so, how often? What is the customer’s diet like? All these things are important factors that contribute to a person’s health. Our goal is to paint a better picture of the customer’s overall health by gathering more information about the person’s lifestyle on a day to day basis, rather than just looking at the person’s previous health information. This will allow our customers, State Farm, to give their customers more accurate life insurance rates and possible discounts depending on how healthy the individual is. Upon completion, our product will provide State Farm customers with a mobile application to track their health information to calculate a corresponding fitness level. This corresponding fitness level will then be used to potentially provide life insurance customers with a reduced life insurance rate. The person’s heart rate, sleep, exercise, driving habits, and location will be tracked by our product.
Background
Before a client’s life insurance premiums are calculated life insurance agents gather a lot of information about the person’s health. Insurance agents will gather some initial information such as the person’s age, gender, and whether the person uses tobacco. Then, they will look at the client’s family medical history. They will also gather more information to determine the individual’s overall health which includes their body mass index and a prescription check. Finally, some insurers will require a physical exam which includes a blood test, blood pressure measurement, and a height and weight verification. The problem with this process is that it does not paint a very accurate picture of the person’s overall health. Even with all this information that insurance agents gather, it does not provide the insurance company with information about the person’s everyday lifestyle. Does the person exercise and how often? What is the person’s diet like? How much sleep does the person get? All these factors contribute to a person’s overall health but are not accessible to the insurance company. With an application to track a person’s sleep, exercise, and diet that will work with a Fitbit the State Farm can take this additional data and give their customers more accurate life insurance rates. Creating an application like this will benefit both State Farm and State Farm’s customers. State Farm will benefit because their insurance agents will not have to collect as much information from clients, which will decrease operational costs for data collection and analysis, which save them time and money. And State Farm’s customers will benefit by being provided with more accurate life insurance rates and possible discounts depending on how healthy the individual is. Say a person has diabetes in their family history, which would count against them and increase their rates, but if the person maintains a healthy diet and exercises regularly State Farm having access to this information would benefit the customer and lower their rates.
For this project, State Farm is our customer that we will be delivering a prototype of the application to. State Farm has had this in their product backlog for a while, however, it is a low priority for them, but they would like to have a prototype of the product. They want us to build a prototype for them, but also want input as to how ethical a product like this would be. As of right now, our team is still working on getting in contact with State Farm. Once we get in contact with them, we will know more about the features that they want to be implemented in this app.
Project Requirements
- The application shall be deployed on mobile device
- The wearable device must be capable of capturing health data
- Wearable device and Mobile device must communicate user data
- Mobile application shall display data captured by a wearable device
- Mobile application shall have a login/security interface
- Mobile application shall correlate given health data with a corresponding fitness level
- Mobile application shall utilize AWS as a means of storing and processing user data
- The application shall detect fraudulent activity
- The system shall be deployed on a Web app
- The website that mirrors Mobile application shall allow users multiple points of entry to their data
System Overview
- Presentation Layer Description
The presentation layer will include the android interface that the user would interact with as well as a web UI for an administrator to interact with. The android UI will be on the user’s phone and will include a log-in screen, screens for the user to view their health data, and a screen for the user to view their fitness score. The presentation layer will also be responsible for capturing the health data from the wearable device, then sending that data to the SQLite database. Another component of the presentation will be the Web UI. The Web UI will be used by system administrators. With the Web UI administrators will be able to log-in and access user data.
- Business Layer Description
The business layer will include AWS, from which we will be using many different services. The business layer will also be where the data analysis is done. There will be a Route53 subsystem, this service will be used to manage all incoming and outgoing traffic requests and will act as DNS Manager. The VPC service will allow a private network to be built, where all the cloud computing resources will be allocated. A load balancer subsystem will also be present to allow AWS ELB to be spawned up into multiple instances and easily route traffic through them. Next, we will have an EC2 subsystem, EC2 is a service provided by AWS for acquiring computing power. AWS EC2 will be used to host our WordPress site and necessary files. The API Gateway subsystem will source a primary point of contact to our VPC through HTTP/HTTPS traffic routed via Route53. The Web App Service will be responsible to allow multiple entry points for users and will also allow admins to check the system status and data analytic. The final subsystem in the business layer is the deep learning subsystem. The deep learning subsystem will be responsible for analyzing the data. There will be three folds to this stack: a CNN based classification Model, a Bi-directional LSTM for forecasting, and a ML based method for fraud detection.
- Data Access Layer Description
The data access layer will store the data that will be collected and provide access to the data. This layer will be composed of 4 subsystems: User, Administrator, Raw data, and Analyzed data. It will interact with both the business layer and the DynamoDB. It will receive data from the business layer and send data to be stored in the DynamoDB. The user subsystem will be where the user credentials are verified. The administrator subsystem will be where the administrator’s credentials are verified. The raw data subsystem will simply communicate the raw data it receives from the business layer to the online database, DynamoDB so that the administrator can view it. Lastly, the analyzed data subsystem is present to communicate the analyzed data from the business layer to the online database, DynamoDB.
Results
Admin Portal, End-to-End ML demo (Video)
Customer UI (Video)
Future Work
- Make the app cross platform compatible.
- Right now the app syncs with Fit bit watch, so make it compatible with other major smart watch like Apple
Project Files
Project Charter (link)
System Requirements Specification (link)
Architectural Design Specification (link)
Detailed Design Specification (link)
Poster (link)
References
“Fitbit Development: Web API.” Fitbit, dev.fitbit.com/build/reference/web-api/.
