In an earlier article, I shared how medicine has used data science for well over 100 years to solve some of the most significant challenges diseases have thrown at us. These early superheroes of medicine attacked the bugs from all sides — locating the transmission sources, helping to create new medicines, and proving the efficacy of treatment programs. Today, we see brand new ways that data science is being leveraged to improve outcomes.
In the fight against the novel coronavirus (COVID-19), there have been a wide swath of use cases being discussed, and in some cases, implemented. Here are a few of the top ones you may have missed.
Direct Medical Use Cases
Artificial Intelligence (AI) Powered Drug Discovery
Recently, Sage Health sponsored an open-source Drug Discovery Competition, where the three top submissions identified potential treatments using machine learning. Several drug discovery companies are also using AI-powered techniques to discover both new and existing molecules that can be used to combat COVID-19. The top existing drug (Remdesivir) is currently being evaluated in several trials. Another open source invitation has been extended by the White House and a coalition of leading research groups that is just getting started as well. If you’re interested, you can learn more here.
Facial Recognition Cameras with Automated Temperature Scans and Mask Detection
In China, officials are taking temperatures of individuals as they enter crowded facilities. Those with elevated temperatures are then taken for advanced screening, potential treatment and isolation. In addition, many cameras can now identify individuals and detect the presence, or lack of, a mask (as required in public places in some countries) to help enforce the quarantine.
Use of Fitbits to Track Disease Spread
In a February 2020 article in the Lancet, researchers questioned whether wearable fitness devices (like a Fitbit) could be used to track the spread of illnesses. While the study was not specific to COVID-19, it showed a strong ability to track disease spread by monitoring and analyzing changes in an individual’s resting heart rate. And because some wearable devices collect additional information such as pulse oximetry, ECG, or even cough recognition, you can imagine how this can be taken further.
Use of Cell Phones to Predict Infection Spread and Contain/Quarantine High Risk Individuals
Korea, China, and Italy have implemented smartphone applications to help manage the quarantine of individuals. In the case of China, cell phone data is reportedly being leveraged to predict where potential outbreaks might happen next. By understanding who has contracted a disease and seeing their prior movement based on their phone’s location services, it’s possible to see who else may have been exposed, as well as in what areas exposure is most likely. Modeling this data, the Chinese government was able to predict which areas (neighborhoods and cities) would be at highest risk, and who would benefit the most from a proactive quarantine. There’s discussion in the U.S. about leveraging this type of data as well.
Rapid Dissemination of Information
While we have seen governments, businesses, and NGOs rapidly collect and share data, there have been instances in which this collaboration seems to be making an especially great impact. In Taiwan, the government quickly made patient traveler history available to hospitals so that risk could be more properly assessed. Other information, like sites that infected people had visited, were posted and those who had visited those locations were asked to self-monitor and self-quarantine as necessary.
AI Diagnostics of CT Scans
AI-based software is being used to interpret CT images of patients’ lungs to look for signs of COVID-19 infection.
Analytics + Business as (Anything But) Usual
Supply Chain Analytics
While medical supply companies and government agencies use population density information as well as current incident rates to predict which hospitals will have the highest demand, companies are similarly looking at their supply chains. Analytically savvy businesses are analyzing goods supplied from high-risk or quarantined areas to quickly identify pre-approved part or material substitutions and activate product or material redesigns. This is occurring not only with direct medical-related supplies, but in every industry segment.
As inventory risks are identified, demand can be shaped with changes to offerings or discounts to help balance inventory. Combined with marketing optimization efforts, this strategy can be a powerful response to keep businesses moving forward during difficult times.
Many companies have transitioned to analytically driven financial forecasting, with some leveraging economic indicators to help provide early warning of significant changes to their sales cycles. These can be challenging when historic precedents do not exist, and it will be interesting to hear how these models perform in the wake of the pandemic.
We have seen companies and healthcare providers leveraging analytics to quickly optimize their scheduling to fit the new demand and rapidly changing constraints. Whether you are a healthcare system identifying high risk patients or re-scheduling elective procedures, or a retail business needing to quickly respond to demand changes, analytics are at the heart of efficient and effective action.
Data Science Gone Awry
Gaming the System: Unintended Analytic Outcome
Following the outbreak of COVID-19, students in China were told to download an app called DingTalk to receive instruction and assignments while in quarantine. Quickly, the kids realized that if they collectively reduced the rating of the app, it would be removed from the app store. On Feb 11, 2020, the app received over 15,000 one-star reviews in a single day, reducing its rating from 4.9 to 1.4 stars — not exactly the goal of a rating system, but a very clever response to the analytic process that ratings are a part of! In an attempt to win back the students, DingTalk issued a humorous apology video. In the end, the app remained running, but it’s always good the think about the consequences of how analytics will be used and if there might be unintended effects.
Whether you’re a hospital looking to quickly leverage data to optimize scheduling, understand the effects of a pandemic on your system, or ensure the right level of inventory for key items, or a business that may be impacted by the societal changes that are occurring, data science can be leveraged to more efficiently respond.
At Alteryx, we see companies in every sector driving double-digit improvements to revenue, operational efficiency, and cost. Perhaps more importantly, we see analytics changing and sometimes even saving lives. With the ability to create new analytics quickly and with self-service tools to deploy them instantly, it’s no surprise to see solutions already in full production in response to recent events.
Chat with us at @alteryx and share any great resources/stories you’ve found.