As coronavirus infections continue to spread, doctors and researchers are looking for new tools that can help them with better and faster diagnosis. Kahun, an Israel-based medtech startup, has released a new COVID-19 tool designed to help doctors make quicker decisions.
The freely available tool is powered by an AI engine, which was fed more than 2,000 papers and articles sourced from the medical library PubMed. It allows doctors and researchers to devise a score from various symptoms to decide if a patient is at high risk of moving to a critical stage.
Kahun’s research team trained the AI model to analyze these papers and give a score of probability as an output. The algorithm uses a technique called differential diagnosis, which lists types of diseases you might have because of your symptoms.
Dr. Michal Tzuchman-Katz, one of the co-founders of Kahun, said that the knowledge graph is updated in real-time as more research related to coronavirus becomes available. She added that the score and the graph are a reflection of how a physician thinks in the real world while treating patients.
Recently, Google released free coronavirus datasets to help come up with new models related to the epidemic ahead. Eitan Ron, another co-founder of the company, said that while these datasets are important for research, Kahun’s tool is more human-friendly:
These data set are information for algorithms not readable by humans. Knowledge related to COVID-19 is provided to physicians and researchers using textual articles.
Kahun knowledge graph provides two major improvements: easy access to actionable evidence-based insights from the ever-growing number of articles published and an AI engine backing up the clinical thought process and decision making.
As researchers gain more knowledge about COVID-19, we’re seeing a bunch of sophisticated AI models emerge. Last month, a team of researchers from the US and China published a paper describing a tool that can outline patients that are vulnerable to developing critical conditions.
You can check out Kahun’s new tool here.
Published April 1, 2020 — 13:52 UTC