AI Challenge + Early Detection of Foodborne Illness Outbreaks
Use Case Description
Foodborne illness outbreaks pose a significant threat to public health and detecting them early can prevent the spread of the disease. Traditional methods of identifying such outbreaks rely on customers reporting their symptoms and complaints directly to health authorities, which can be time-consuming and unreliable. However, social media platforms like Facebook, Google Reviews, Yelp, and others can serve as a rich source of data for public health research. By analyzing social media posts that mention symptoms or specific foods, researchers can identify potential outbreaks of foodborne illnesses and track their spread.
To extract useful information from these social media posts, we can use NLP techniques such as Text Classification and Named Entity Recognition (NER). These techniques can identify relevant symptoms, food items, and the location of the post, which can be enriched further via structured data extraction using RPA.
My proposed approach for this Intelligent Automation use case is as follows:
To automate the process of extracting reviews and feedback from popular restaurant review sites, such as Facebook, Google Reviews, iwaspoisoned.com, OpenTable, and Yelp, we can use UiPath RPA with Python libraries such as Scrapy and Selenium. By automating the data extraction process from multiple sources, we can gather more data quickly. While UiPath RPA alone can do the extraction, but based on my research and testing, use of Python libraries helps in faster extraction and is less dependent on UI changes of websites. Additionally, the Robot can extract structured data associated with the review, such as the restaurant name, address, website, review date, and time.
Once the reviews are extracted, the Robot can call the ML models in AI Center to perform Text Classification and NER on each of the reviews. To manage a large volume of reviews, we can use sentiment analysis to narrow down to negative reviews before applying Text Classification and NER. This speeds up the execution and consumes fewer resources (AI Units). We can either use the default packages in AI Center for Sentiment Analysis, Text Classification, and NER, or use Communication Mining (Re:Infer) to get better results. The choice of ML models would depend on which models give better results.
Overall, the use of NLP techniques and RPA automation can provide an effective way of detecting potential outbreaks of foodborne illnesses from social media posts, allowing researchers and health authorities to respond more quickly to prevent the spread of the disease.
While the primary customer for this use case would be the health authorities, the use case can be easily extended for other stakeholders such as restaurants, malls, etc. where we can classify the reviews across other categories such as food quality, service quality, pricing, cleanliness, location, customer satisfaction, etc.
AS-IS WORKFLOW, TO-BE WORKFLOW
Other information about the use case
Industry categories for this use case: Customer Service, Healthcare Pharma, Public Sector
Skill level required: Advanced
UiPath Products that were used: UiPath Studio, UiPath Action Center, UiPath AI Center, UiPath Communications Mining, UiPath Orchestrator
Other applications that were used: Python libraries, browser, Excel
Other resources: This automation provides savings of 240 man-days per year. In addition, it provides higher completeness of retrieving the relevant cases, as doing it manually, users would often be not as comprehensive as the automation. This also provides compiled data for further research.
What is the top ROI driver for this use case?: Accelerate growth and operational efficiency