The pioneering collaboration between Sheba and Open AI: the possibilities and the trap

At the same time, the number of doctors, nurses and hospital beds per capita is relatively low. And on paper, AI can help with that. It cannot build a new department, place an additional hospital bed or train a specialist, but doctors and nurses spend a significant part of their time on documentation, summarizing cases, locating information and entering data. Shortening a few minutes in each medical encounter may add up, multiplied by millions of visits, to a significant saving in working hours.

However, hours saved are not automatically recorded as budget savings. Amr cites a study published in JAMA Network Open and estimated that AI tools for documentation may add to a doctor an output worth approximately $3,044 per year. However, the license cost of these tools usually ranges from $300 to $500 per month per doctor ($3,600 to $6,000 per year) – even before including the cost of integration calculations, information security, training and ongoing maintenance. The bottom line result is often a net loss.

“As long as the work process itself does not change to convert the hours saved into measurable output, technology is not an engine of growth, it is simply another item of expense,” says Amar.

The problem is particularly evident in large organizations that operate using old systems. A model can summarize a document in seconds, but if the worker is still required to manually copy the result, re-enter data and switch between several programs, most of the savings are gone.

“The model is only one layer, the real challenge begins with connecting it to the existing information and systems, and with the redesign of the work process. Only then can you check if an action that took ten minutes now takes two, and if the eight minutes that were freed up really generate value elsewhere.”

Therefore, he concludes, the economic test of the cooperation between Sheba and OpenAI will not be the number of users or the scope of operations that the system knows how to perform. It will be tested in the decrease in working time, in the increase in the number of patients that the same team is able to serve and in a comparison between the increase in productivity and the full implementation costs.

Kzor, Sheba Hospital Tel Hashomer is expected to be the first hospital in Israel and the first outside the United States where artificial intelligence tools will be used by the medical staff. The system integrated with artificial intelligence is expected to start operating at the beginning of the next school year, on September 1, after the implementation process and safety tests are completed.