An alert system for possible critical diseases for a medical startup

This case study focuses on the features implemented in a medical app for a healthcare startup. The app allows users to track their health, record chronic pains and monitor their progress and treatment. It also offers a service for medical institutes, enabling doctors to access patient records and contribute their own information.
The task was to develop an automatic alert system to detect critical illnesses based on the data users enter in the application. Users enter hundreds of thousands records of detailed information per year (in 2024 there were approximately 1 million records) about what pain they are experiencing, how they are feeling, what medicines they are taking, and under what conditions, such as exercising or eating and so on. Based on this data, it is possible to predict whether there are serious problems that require immediate medical attention. This avoids complex and expensive procedures, including those requiring long patient recovery times.

Project Overview:

• Location: Canada
• Product: Website and mobile app
• Technologies: Node.js, Express.js, TypeScript, Drupal, Symfony
• Team: Back-end Developer, Front-end Developer
• Timeline: 2016 – 2017

Regrettably, we cannot disclose the application's name, any app screenshots, and links to the product due to a non-disclosure agreement (NDA).

Initially, this alert system was designed exclusively for medical organizations to monitor inpatients and patients treated at home. It effectively notified doctors of potential serious illnesses and complications, allowing them to catch diseases in their early stages and avoid severe consequences for patients. The system identifies and analyzes user data according to predefined criteria, triggering alerts to medical professionals when specific events occur.

The criteria were based on matching the conditions of multiple patient records, which meant that the system had to analyze thousands of patients’ records. А configurable record processing system was implemented, where all patient records are processed and analyzed, and all necessary data for each alert is calculated. The results obtained are stored in a database, from which medical professionals can quickly obtain information on whether a particular alert has worked for a patient or not. This system is based on a queue and this ensures that all data will be processed, and if a technical error occurs, all data processing would be paused and support engineers will be notified about the problem.
As the number of complications and diseases defined grew, the system was upgraded to send over 30 different types of alerts: from severe pain alerts and changes in them to medication side effects. The system has further evolved to include notifications for patients about whether they add new records to the app or not, and now the alerts system is an integral part of an overall notification system (similar to Duolingo, but not as persistent as Duolingo). The app's customers are also medical organizations and insurance companies, and keeping user information secure is important to them, so the company that developed the app passed a risk assessment and received SOC2 Type 2 certification.

The implementation of the alerts system has led to significant results, as outlined below:

• A monitoring system was created for medical organizations. Its main task was to save patients' lives and reduce the risks of serious complications and disease courses. The system effectively monitored patients' health and optimized treatment.
• The alert system was profitable for health insurance companies because their clients underwent simpler and cheaper procedures in the early stages of illness, saving money. Also, insurance companies did not lose customers because their customers did not die from misdiagnosis, late treatment or procedures.
• Anonymous medical data is valuable for scientific research. The introduction of a data analysis and storage system made it possible for the first time to obtain a large and detailed dataset updated in real time by patients themselves. Scientists and doctors can create research papers and articles based on this data and raise citations and increase the prestige and popularity of the clinic.
• The system's efficiency and advantages make it possible to find new clients among medical organizations of other specialties, such as psychologists and other target groups with sufficient budget.
• The system became the basis for patient notifications and other integrations related to the application and user inputs. This functionality is easy to expand and scale.