The Alarm Dialog Management (ADaM) project team has expanded the capacity of Public Invention’s existing alarm devices, the Krake and GPAD, by creating a system that prioritizes alarms for a given user. This places some of the decision making on the design algorithm and off the user, thus helping them avoid alarm fatigue.
Public Invention Founder Robert Read is the Invention Coach currently heading the ADaM. Also working on the project is software engineer Saicharan Vishwanatha, who developed the ADaM’s algorithm using JavaScript. In an interview, Vishwanatha shared how his educational and professional background supported the project’s development.
“I did my undergrad in India in computer science. I also worked at TCS for more than two years as a software engineer, where I [worked] on Java[Script] backend and front end technologies,” Vishwanatha said. “After working with TCS for two years, I felt like I wanted to study more. I came to the U.S. to do a Masters…at Colorado State University in computer science.”
Armed with this experience, Vishwanatha joined Public Invention and began work on ADaM’s algorithm. He shared the story of its development, the hurdles and successes it faced, and his aspirations for the project’s final form.
“The main goal for the ADaM project is to solve a real pain point in high-noise environments, like in a hospital’s ICU, where dozens of alarms fire at once and overwhelm human responders.” – Saicharan Vishwanatha
The ADaM’s Development
Defining the End Goal
ADaM is a medical alarm-based project that not only makes an alarm more effective, but also more efficient. It accomplishes this aim through utilizing a unique, open-source system known as the Dialog, which manages both alarm priorities and human responsiveness. The goal of ADaM is to make a reusable system that manages alarms based on human interaction with overlapping alarm conditions.
ADaM began with the goal to minimize alarm fatigue for those working in medical settings, like the ICU. There was a need for a device that would prioritize alarms based on their urgency and reduce those with low urgency. This would allow the user to focus on high-priority events, and avoid becoming overwhelmed and tuning out alarms haphazardly. At the same time, the low-priority alarms would need to temporarily silence, but not dismiss altogether. Otherwise, the user could risk ignoring them, and thus undermine the alarm’s usefulness. Vishwanatha stated this goal succinctly.
“The main goal for the ADaM project is to solve a real pain point in high-noise environments, like in a hospital’s ICU, where dozens of alarms fire at once and overwhelm human responders. The goal is to create a system that intelligently manages how the alarms are presented, prioritized or muted, and resolved, so that the operators get the right information at the right time without missing any crucial events,” he said.
The Algorithm
Vishwanatha utilized his background in JavaScript to code the system that would make these priority decisions. However, despite his previous experience, ADaM presented its own set of difficulties.
“The real challenge was designing interactions that reduce human load when a new alarm lights during an active dialog. The system initially interrupted [the alarms] too aggressively, and fixing this required writing a stable arbitration layer that squeezes new alarms or re-evaluates priorities. This decides whether to interrupt or defer.”
There were also key moments of progress on the project, which Vishwanatha readily described.
“I hard-coded alarm logic into a structured configuration format. This made our domain onboarding much easier and validated the project’s main, core vision of configured dummy code,” Vishwanatha said. “Additionally, after building the priority and suppression logic, simultaneous tests showed a clear drop in low-value alerts.”
This emulated more of the prioritization model he hoped to incorporate in the system architecture.
Publication and Hardware
Recently, Vishwanatha made a mathematical model of ADaM using a Node.js environment. He and Read performed a Monte Carlo Simulation to test the probability of different outcomes during the algorithm’s use. The project team has now begun work on a paper that will share their findings from this simulation, and they continue to test ADaM in JavaScript and Python environments.
“The ADaM project and any other project in Public Invention…needs to get more PR to visualize what we have done, what’s our goal, and how it’s going to help [others].” – Saicharan Vishwanatha
Although users can currently access a version of the ADaM online, Vishwanatha emphasized that the team plans to add hardware to the device once the software is fully developed. That way, those who need such a device can have a completed product, rather than needing to move the software into a pre-existing alarm system.
“If everything goes well, we would look into the hardware too,” Vishwanatha reflected. “The main goal is a combination of hardware and software.”
Spread the Word: Open-Source Projects and Mission
Beyond these aspirations for the ADaM and its current progress, Vishwanatha emphasized the wider need for spreading the word about open-source projects and the open-source community. This garners necessary support for open-source efforts, and spreads awareness about opportunities that organizations, like Public Invention, provide.
“The ADaM project and any other project in Public Invention,” Vishwanatha said, “needs to get more PR to visualize what we have done, what’s our goal, and how it’s going to help [others].”
ADaM and its Inventor Vishwanatha highlight the opportunity open-source projects provide; to bring one’s experience to a topic that will positively impact others. Each of Public Invention’s alarm devices, from the GPAD project to ADaM, show new opportunities for alarm systems that will help medical professionals and those they serve. Most of all, Vishwanatha’s contributions to the project show the impact inventors can have on the future of open-source and medical software. Ultimately, time will only tell what the full impact of the ADaM will be on the field of alarm management.