Using AI as a tool for infectious disease surveillance

Photo Credit: National Cancer Institute/Unsplash

Recently, sites like ChatGPT have become hubs of information via chatbots, and the topic of artificial intelligence (AI) is on the rise. Unsurprisingly, the potential for his AI assistants like ChatGPT to play a role in health through medicine and public health is becoming more and more debated, and the larger concept is being debated for its pros and cons.

In recent studies, JAMA Internal Medicine Rated Such tools in medicine “compare written responses from” doctor From ChatGPT to real world health questions.A panel of qualified health care professionals is preferred ChatGPT responses 79% ChatGPT responses were rated as higher quality and more empathetic. ”

But in this context, given the experience of the last three years, how can AI be leveraged for something so important and relevant: infectious disease surveillance?

Recent Publication in the New England Journal of Medicine, posed exactly this question. We look at the success of various AI applications ranging from early warning systems, epidemiological forecasting, and even resource allocation. The authors sought to evaluate how AI could support disease surveillance through early warning tools, distinguish between different epidemics, and how such tools could track the source of outbreaks. .

The authors provided an in-depth review of various AI capabilities in disease surveillance, from risk assessment to source identification to hotspot detection. This not only strengthens our response to outbreaks, but also a stronger preparedness system. A particularly interesting example provided was source detection, citing work from the University of Pittsburgh that developed the EDS-HAT. EDS-HAT is an enhanced detection system for healthcare-related transmissions that combines whole-genome surveillance sequencing and machine learning to extract data from patients. Electronic medical records during outbreaks.

The authors stated, “The algorithm combines EMR data from patients known to be infected by the same outbreak (case) and EMR data from other patients in the hospital (controls used to establish baselines). ) were trained by a case-control method that analyzed level of exposure relevance). This form of learning prompted the algorithm to identify her EMR similarities (procedure, clinician, room, etc.) in cases with related infections. Analysis of EDS-HAT reveals that EMR-based real-time machine learning combined with whole-genome sequencing could prevent up to 40% of hospital-acquired infections in nine surveyed locations, potentially saving costs Did. ”

What makes something like this so exciting? Ability to identify outbreaks we may have missed, such as drug-resistant bacterial infections in two of her patients who underwent the same type of imaging tests by the same technician. The authors shared another example. “In another example, Pseudomonas aeruginosa An outbreak among 6 patients in multiple hospital wards over a period of 7 months was missed due to large distances in time and space. Genomic surveillance suggested that these cases were all related, and machine learning algorithms identified a contaminated gastrocamera as a likely source and an easy target for intervention. . ”

These are just a few examples of a single category of applications reviewed that ultimately demonstrate AI’s increasing ability to support outbreak response and rapid surveillance efforts. The authors mention some challenges that need to be addressed, such as the recognition that these tools should serve as supplements and tools rather than a replacement for traditional high-quality surveillance, but AI There is also a recognition that there is no substitute for coordination between partners across jurisdictions and functions.

However, we have the goal of building a stronger analytical and collaborative surveillance network, which could improve this ability. After all, emerging technologies such as AI have a role to play in disease surveillance and epidemic science, and now is the time to build infrastructure and integrate them with existing systems.

look at us recent interview Discuss AI with physicians who are chief medical information officers in major health systems, including how AI might be applied to journal articles and how clinicians are already using AI doing.