I have read your project proposal and find it to be very intriguing. From my understanding, you hope to build a model to identify patients with COVID based on the COVID-like symptoms they have reported. The dataset you will be using is collected through an off-platform survey launched by a joint effort of CMU, UMD, and Facebook.
There are 3 things I loved about your proposal:
Informative dataset: it consists appropriate amount of records well distributed amongst different countries. Each record consist of abundant information about the patients like their geographical location, demographical information, symptoms, and etc
Clear approach: you already have a few statistical modeling methods in mind to tackle this complex problem which boosted my confidence in this project
Bring positive impact: your project will attempt to identify patients as soon as possible to prevent further spreading. This is really meaningful as the whole world has been suffering from COVID for more than 1 year.
There are 3 things that make me hesitate:
Biases in the dataset: since the data is collected through the method of self-reporting, there could exist some biases/uncertainty. For instance, the level of how severe their symptom could variate a lot
Short timespan: the dataset only spans less than 1.5 years (Apr. 2020 - Oct. 2021). Will this be enough information for you to capture the time-trend (new variates of COVID & pre to post-vaccine phase)? How would you validate that?
Usability: as more and more covid testing methods and options, especially self-testing, are available, what are some reasons that people or organizations would prefer your model over COVID testing?
Overall, I loved this project idea and looking forward to hearing updates on this.
Dear Valued Employees,
I have read your project proposal and find it to be very intriguing. From my understanding, you hope to build a model to identify patients with COVID based on the COVID-like symptoms they have reported. The dataset you will be using is collected through an off-platform survey launched by a joint effort of CMU, UMD, and Facebook.
There are 3 things I loved about your proposal: Informative dataset: it consists appropriate amount of records well distributed amongst different countries. Each record consist of abundant information about the patients like their geographical location, demographical information, symptoms, and etc Clear approach: you already have a few statistical modeling methods in mind to tackle this complex problem which boosted my confidence in this project Bring positive impact: your project will attempt to identify patients as soon as possible to prevent further spreading. This is really meaningful as the whole world has been suffering from COVID for more than 1 year.
There are 3 things that make me hesitate: Biases in the dataset: since the data is collected through the method of self-reporting, there could exist some biases/uncertainty. For instance, the level of how severe their symptom could variate a lot Short timespan: the dataset only spans less than 1.5 years (Apr. 2020 - Oct. 2021). Will this be enough information for you to capture the time-trend (new variates of COVID & pre to post-vaccine phase)? How would you validate that? Usability: as more and more covid testing methods and options, especially self-testing, are available, what are some reasons that people or organizations would prefer your model over COVID testing?
Overall, I loved this project idea and looking forward to hearing updates on this.
Sincerely,
Your Manager