Implementing IA in operating rooms at public hospital would require a $9 billion investment


Researchers at the National Center for Artificial Intelligence are currently working on a project that would make it possible to implement state-of-the-art technological tools to manage tertiary healthcare system operations.

Researchers from the National Center for Artificial Intelligence (Cenia) estimated that the country would need an investment of $9 billion to support, through “AI” tools, the management of operating rooms in high complexity hospitals in Chile. Together with the quantification, the specialists have drawn up a project that would enable these tools to be used in one facility or in the entire public health system. Dr. Cristian Buc, researcher at Cenia and who leads the team that is currently exploring opportunities to implement the initiative in a real environment, emphasizes that the investment represents a minimal cost compared to the potential benefits in terms of efficiency and productivity in the surgical areas of the 16 high complexity hospitals operating in all the regions of the country.

“This is a problem that has been going on in Chile for years. If we look at the metrics of how many resources are invested in public health and its efficiency, it is basically an increase in money while the performance remains flat. That is to say, more and more resources are injected, but the frequency of operations, and their associated waiting list, remains the same,” says the specialist from the agency dependent on the National Agency for Research and Development (ANID).

According to a recent report by the National Productivity Commission, the management of operating rooms is one of the main challenges for public health and one of the most complex causes for resolving waiting lists. The report also highlights the urgency to incorporate new technological solutions into the system in order to strengthen its capacity to provide care. Artificial intelligence, the mainstay technology of the fourth industrial revolution, is expected to consolidate as a key support technology for clinical teams throughout the care process. On the one hand, by contributing to the diagnosis and early detection of diseases through the analysis of large amounts of health data such as medical history, medical images or clinical test results. On the other hand, it enables a more agile management of processes that in the past were an additional burden for professionals. It is therefore expected that the increasingly widespread adoption of new solutions will enable healthcare professionals to focus on more important tasks, such as spending more time with their patients, while improving the efficiency of their work.

A report by Universidad San Sebastián’s IPSUSS, using publicly available data provided by the Ministry of Health, estimated that if the local health system had used all its installed capacity between 2017 and 2019, annual surgeries would have practically doubled, reaching one million every twelve months. During these years, only 52% of available time slots were used, which could even cover the entire waiting list (around 270,000 surgeries).

Digital twins

The Universidad San Sebastián study pointed out that the main causes for the low use of operating rooms are as follows: 14% of the available time slots are lost due to lack of equipment, personnel or both (this equals 150 thousand major surgeries per year); the other operating rooms are used 70% of the time, which means a loss for the system of another 450 thousand surgeries, that are affected by the low compliance with the schedules.

The project designed by the specialists of the National Center for Artificial Intelligence has two levels. One, the creative level, is based on a tool called “digital twins,” and its adoption aims to model the operating room and the different players in this environment, as well as the different parameters involved in its management. “A digital twin is the representation of a physical object in a virtual world. It can be, for example, an airplane turbine. I model the turbine in a virtual space and by that I can model how this turbine works under different parameters, seeing the effect that these factors could have on the turbine`s lifecycle,” explains the researcher from the Chilean technology center.

According to Cristián Buc, this operates as a sort of machine of the future to make better decisions, that is, to move a group of professionals from one operating room to another, to call another anesthesiologist in case an operation has been extended by another hour, etc. Basically, to foresee what happens if this and that happens, maximizing the use of operating rooms. What happens, for example, if a specialist is stuck on a traffic jam from one hospital to another?

Incorporating technology could ensure that even with these contingencies, the surgeries scheduled will not be lost. The data to build these models will come from sensors installed in the operating room and equipment, and geolocation systems for professionals, allowing the technology to work in real time. “With the digital twins we can simulate at a speed that is incredibly faster than our real time. In a fraction of seconds, with the current state of the room, we can manage so that this time scheduled is not lost thanks to better decision making. The idea is precisely to maximize the use of operating rooms.”

Neural networks

Currently, the average waiting time for surgery in the Chilean public system is around 525 days. Several reports from organizations that have explored the phenomenon state that if the 52% of hours actually used were to reach a figure close to 70%, the impact of the problem could be much less. For this reason, the second level on which Chilean researchers have focused is the proactive one, in which they use optimization and deep learning models to create an adequate organization for the management of the operating rooms. In the event that this sequence breaks down, the “digital twins” come into action, providing recommendations to prevent these surgeries from being lost in the face of unavoidable system contingencies.

The design of an optimization model based on neural networks will require obtaining quality data to estimate the key parameters for its operation. In fact, both mechanisms (proactive and reactive approaches) work as a perfect complement to maximize available resources and prevent patients from continuing to wait when resources are available to attend to their emergencies.

Cenia researchers hope to have support from Minsal so that a project based on the proactive approach (spreadsheets based on optimized models) can be adopted in a local facility. “Neural network models can be useful in estimating surgery times and they already exist. To implement them, we just need to make parameter changes to see which of all the architectures is best for their operation,” noted the leader of this project.

In 2021, waiting lists for care in the Chilean healthcare system caused up to 30 thousand deaths, according to official figures. It is estimated that, in total, the number of users waiting for an appointment for specialists, GES pathologies or surgeries is 1.7 million people. Just over 15% of this percentage corresponds to the unavailability of operating rooms. “This is a gigantic number and causes an enormous number of deaths in the country. We hope that the technological solutions, which could make an important contribution to facing the great challenges of the health system, do not clash with the reality of the country, which will undoubtedly depend on good decisions by the authorities,” reflects the Cenia specialist.

Data, information and decisions

According to data from the Ministry of Health, 90% of the causes for the suspension of surgeries are avoidable. Previous studies for the same period found an average delay of 40 minutes in the beginning of the surgery and an early end of 1 hour and 42 minutes. For Cenia specialists, a large part of the problem is a consequence of the lack of flexibility in the management of physical, economic and human resources. This is also part of the recommendations made by the USS study.

“The efficiency problem arises from how the times are organized. What many entities in Chile do is that they create spreadsheets with hours for interventions based on the available personnel. But what happens if the hours are CANCELLED or can it be predicted? Well, there is no flexible process to use those cancelled slots, which are going to be left there completely available. It’s a huge opportunity cost to the system,” Buc said.

The most crucial part of the project is the collection of data and the availability of this information for everyone to access. The idea is that this spreadsheet can be viewed by all players, incorporating tools for geolocation of agents’ functions in care centers and calendar notifications similar to those available in e-mails.

For Cenia’s group of experts, the aim is to help the public sector strengthen its technological capabilities for the benefit of users. “The hospital of the future is a digital hospital at all levels. Our project in particular addresses the problem of operating rooms, but we are already looking in Europe at tools for scheduling appointments, and in the United States a digital hospital is being planned for all processes, which requires an incredible capacity to obtain data to generate models. We know that in the public sector we have a digitalization gap.” “We hope to try to contribute with this type of projects,” concluded Dr. Cristián Buc.


By Luis Francisco Sandoval. Agencia Inés Llambías Comunicaciones.






Noticias Recientes