S Fund: Need-Based Hospital Support

Inspiration

Our mission is to provide direct and expeditious aid to under-resourced hospitals. The S Fund is a medical equipment marketplace that algorithmically provides subsidies based on need. The current disunified system of hospital subsidies is not capable of quickly adapting to public health disasters and can lead to dangerous imbalances in resource distribution. This systematic flaw notably inhibited the urgent treatment of individuals during the COVID-19 pandemic when states were forced into bidding wars for sparse medical equipment. As Senator Chris Murphy explained the situation, "Every state, major city, and territory, and thousands of hospitals, are being forced into a bidding war, encouraging price gouging and hoarding.” Corporations have only exacerbated this issue by disproportionately raising prices at higher profit margins (the biggest culprits are comically recognized with Shkreli Awards to highlight those who profited off of the pandemic and healthcare in general). Consequently, under-resourced hospitals with the greatest need have the least economic power in acquiring crucial medical equipment. The S Fund addresses this market failure by using data science and economic principles to efficiently distribute government and donor funding.

What it does

Hospitals that register with the S Fund are connected with relevant suppliers and have financial aid packages that are tailored to their personal needs in a given market state. The heart of our model is our analytical tool for calculating economic need: the S Coefficient. The S Coefficient derives the optimal allocation of a “joint private and government donations fund” to financially assist hospitals in need of specific resources.

How we built it

The Mathematical Formula for the S coefficient: The heart of our model is our analytical tool for calculating economic need: the S Coefficient. The S Coefficient derives the optimal allocation for a joint private and government donations fund to financially assist hospitals in need of specific resources. We derived a formula to calculate the “hospital_need_index”, a metric measuring the degree of urgency with which a hospital needs a particular good. We considered 2 variables: a hospital’s net worth divided by the number of patients (a proxy for a hospital’s ability to pay for patient resources), and the quantity a hospital demands of a product divided by the days until the supplies are needed (a proxy for the hospital's urgency for the supplies). By multiplying these 2 variables, we obtained an easily quantifiable metric for the “hospital_need_index.” The hospital_need_index is multiplied by the average of the prices set by producers in a market (this operation normalizes the hospital_need_index for a given product). Thus, the numerator of the S coefficient represents the relative monetary value of a hospital's need for a given product). The relative monetary value of the hospitals is then divided by the sum of all hospital_need_indicies and their corresponding prices to derive the proportion of total need in the market that the hospital’s supply request occupies. The S coefficient is then used in backend calculations to compute the personalized discounts of a given hospital request at a given time.

Website: Using HTML and CSS, we built and designed a custom website where hospitals could send an “equipment request” form and search for suppliers, while suppliers could offer equipment for sale to hospitals. We included custom HTML forms and searchable and filterable HTML tables. We used MongoDB for backend development and integration with the data obtained through data scraping.

Data Scraping and Manipulation: From the data we obtained, we first found a database of quarterly earnings through a Californian database of different hospitals. Afterward, we imported the CSV file into Google CoLab and did multiple data manipulation techniques through ‘pandas’ and ‘scikit-learn’ in order to extract the data we wanted, along with necessary data calculations that we introduced using existing data. We compiled everything into a singular data frame which we exported as .json and eventually stored the database in MongoDB for backend development and integration. We did this for multiple data sets including those of quarterly earnings in order to figure out the S coefficient, the actual marketplace, data frames, and tables.

Challenges we ran into

The biggest challenges we faced as a team occurred through the execution of our plan. Our team members were constantly rearranging our formula for our S Coefficient while others were persevering through a new coding language. All in all, these slight setbacks did not prevent us from developing a cohesive data model.

Accomplishments that we're proud of

We are extremely happy with the work our team completed to address a real-world challenge. Our team members contributed an array of interdisciplinary skills from economics, mathematics, data science, and graphic design to assemble a unique solution to an important problem.

This ultimately made for a thought-provoking experience, highlighting the importance of an innovative transdisciplinary approach when tackling global problems.

What we learned

As a team, we are very proud of what we accomplished in such a short time. We learned the foundations of both HTML and CSS coding through front-end development. This included learning how to create forms, using buttons to hyperlink between pages of our website, and coupling our iterations with complex database systems.

What's next for S Fund: Need-Based Hospital Support

The S Fund has great potential for broader impact and expansion because it is highly scalable and can be integrated into existing healthcare systems. The S Fund enterprise plans to launch the first version domestically as a non-profit organization with a free enrollment of hospitals and suppliers. Approximately 3% of the S Fund investments will be allotted to outreach, marketing, and digital upkeep, the remainder will be put directly into the allocation algorithm for redistribution. Our initial objective is to create 50 hospital profiles in the first month and 500 after a year to establish a statistically significant and diverse pool of data research and development. Our team will reach out to philanthropic organizations and government offices for initial capital investment. Once a solid database has been established, we will attempt to license our algorithmic process and user interface with a large medical donations organization such as the Red Cross and the World Bank.

The S Fund will equitably distribute subsidies to hospitals to maximize the return on investment of donor funds. Deploying this system of allocation will make the economy of medical supplies more nimble in the face of unforeseen disasters and reduce the superfluous consequences that result from imbalances in resource distribution.