The rapid spread of COVID-19 has critically overwhelmed health institutions around the world, a problem clearly evident in the United States. Amid the COVID-19 crisis, healthcare institutions and government leaders face unthinkable challenges, including a shortage of supplies that threaten both patient care and the safety of healthcare workers. The incessant volume of patients and the high acuity of patients have exceeded the bed and staff capacity of many hospitals. In these difficult times, the need for effective inter-hospital collaboration is increasingly clear.
There are initiatives connecting hospitals with providers and initiatives documenting equipment stocks, but as Dr. Daniel M. Horn, a physician at Massachusetts General Hospital in Boston, urged in his New York Times op-ed, "Great technology You need to quickly build and scale a national cloud-based ventilator surveillance platform that will track ICU capacity of individual hospitals and supply of ventilators across the country in real time, which platform, which Silicon Valley could build and FEMA could use, would allow hospitals across the country to report daily on the ICU bed status and supply of its ventilator, in an unprecedented data sharing initiative. " LifeMed plans to update this platform.
LifeMed will create a near real-time dashboard display showing hospital supply needs to accelerate the distribution of vital hospital supplies and equipment. Our mission is to create a network infrastructure that enables healthcare systems to share data efficiently and collaborate effectively.
How I built it
Built with express.js heroku mapbox mongodb node.js pitón reaccionar
Challenges I ran into
The main limitation of our model is the limited number of cases, as we apply our model to 58 different counties within New York State only. To focus our model on attributes and their relative effects on COVID-19 propagation, we used the Decision Tree as a regression method (Python Scikit-Learn implementation). Decision trees (DT) are a well-known and widely used supervised classification and regression technique. DTs extract decision rules from the training characteristics that would be used to make predictions about new data. These decision rules can be described internally as a set of if-else steps.
The main advantage of decision trees is the simplicity of the model and its white box nature (i.e. precise decision rules can be extracted and displayed). DTs do not require a relatively large data set to function well, compared to other techniques such as neural networks and the support vector machine.
Achievements I am proud of
There is a real need for this app in the field as soon as possible, as keeping people safe should be one of our top priorities in the war against Covid-19. Therefore, we are proud to focus our effort on such an important and urgent need.
What I learned
In this study, we used decision trees to predict the next-day percentage increase in COVID-19 cases at the county level within New York. The results show that sociodemographic data influences the spread of COVID-19 given the promising performance of ML even with a very limited data set. Therefore, the correlation between community traits and COVID-19 cases appears to be valuable in predicting the expected supply needs of a hospital, resulting in improved patient care delivery and safety. of health workers.
What's next for LifeMed?
The main objective of this attempt is to show that it is possible to correlate sociodemographic traits with the spread of COVID-19. Additionally, this demonstration serves as initial proof that machine learning can be used to predict future numbers of new COVID-19 cases. We are limited in the conclusions drawn from the results of this model given the restricted data used and the geographic limitation (only within New York). Also, the number of positive COVID-19 cases depends on the number of COVID-19 tests that are performed. In addition, time series forecasting could be implemented when time series data is available.
To further improve this project, additional effort is needed to:
Collect data for all counties in the United States Extract test rate by county Extract more daily amount of COVID-19 from cases Extract the hospitalization rate of patients with COVID-19 With the extracted relative data, multiple forecasting attempts can be made to predict:
COVID-19 hospitalization rate Death rate Long-term infection rate Length of stay of patients with COVID-19 Hospital fan needed PPE need per hospital
Try It out
express.js, heroku, mapbox, mongodb, node.js, piton, reaccionar