Taking Surgery to the Next Level with Machine Learning

We are not at the point where human surgeons are no longer needed, although robots have been used in surgery rooms for over 30 years. The difference is that recent advancements achieved by machine learning consultants can give these tools more freedom as well as more responsibility. This is good news when it comes to high-precision and time-consuming operations but it still raises significant concerns related to malpractice risks.  

Although the technology is not yet perfect, there is a clear upward trend in this area. MarketsAndMarkets projects an annual growth rate of 10.4%, from $3.9 billion in 2018 to $6.5 billion in 2023. 

There are many ways of how machine learning could improve surgical outcomes. These include automating suturing, evaluating surgical performance, helping young surgeons develop their skills, and enhancing surgical workflows on the whole. 

Robotic Surgery and Better Surgical Robots

The famous da Vinci system has been in use for almost two decades, but there are new tools for precision surgery. 

One of the most delicate acts is suturing. It requires precision and accuracy, is time-consuming, and runs a high risk of being affected by human fatigue. This makes it a perfect candidate for automation with robots. 

Until now, surgical robots have been performing tasks and following precise instructions under human supervision, being simple extensions of the human hands. Current research aims to change this and make robots fully autonomous. 

A good example is the STAR robot that tackles the challenges of working with soft tissue. For this robot to operate, a team of surgeons just needs to set the parameters including suture size, tension of the material used to close the suture, and more. 

Another application for surgical robots is high-precision operations such as eye surgery that requires extreme precision, sometimes in terms of microns.  

The advantage of robotic AI here is that it can self-learn. This means learning not only from initial training data sets, but also during operation in real-life conditions.  

Evaluating and Enhancing Surgical Skills

Every time we face of the need to choose a surgeon, we accept nothing less than the best one. However, what is an objective way to evaluate a surgeon’s skills? 

Until now, the evaluation of young surgeons was an empiric process performed by more experienced doctors. The advancements of machine learning allow making this process more rigorous and easily quantifiable. Possible KPIs for a surgeon’s performance include:

  • time to complete a given task 
  • typology of the chosen sutures 
  • perception of tissue depth
  • operating speed without affecting accuracy
  • smoothness of hand motion during surgery

Good news is that through machine learning, there is the possibility to evaluate a surgeon’s skill level in about 85% of cases. The results could also help surgeons to bridge their skill gaps. This is similar to training airplane pilots, it’s just that the stakes are usually higher, as is the level of details. 

It is the perspective that could scare some of aspiring surgeons while motivating others. Yet, it gives patients the option of an informed choice and lead surgeons the possibility of creating skilled teams.

Better Procedures and Workflows 

Although there are strictly regulated procedures to ensure the best outcomes of surgery, there are still risks of complications and hence mounting hospitalization costs. All industries, from hospitality to civil aviation, use checklists to ensure compliance and mitigate risks, and surgery rooms are no different. 

A system called IDEAL-X uses machine learning to extract structured data from medical records in real time. The results can be used to file compliance reports, keep patients’ conditions under control, and improve workflows, giving doctors and nurses more time to spend taking care of patients instead of doing paperwork. 

Machine learning can also be used to develop procedures and best practices for specific situations.

In the case of burn surgery, there is the need to assess the affected area first, a procedure that was done primarily by gross approximations before. The results were used to evaluate threats to patients’ lives. Currently, machine learning can take the guessing game out of the equation, by assessing not only the affected percentage of the body but also the degree of the burn, computing risks on the spot. 

The Current State and Future Developments 

Right now, there are over 5,000 robots in use, all with varying degrees of autonomy. Some leave decision-making to the surgeon and only assist with fine-tuning like stabilization, while other systems have more free will. These are used in neurosurgery and eye surgery for their ability to perform finest motions while following presets provided by surgeons.

Even surgeries like hip replacement can be performed by robots, which operate in a similar way to CNC machines. With a 3D plan to follow, they can tirelessly complete their tasks with high precision. 

The future belongs to systems with increased autonomy. These robots have learned from extensive medical databases and can combine the know-how of multiple generations of surgeons into new approaches. Of course, the licensing process for these tools will be long, and the tools themselves need to go through extensive testing before they are fully accepted. 

In the near future, we can expect minimal invasive AI-powered systems that would require just slight supervision from a lead surgeon.

To Top

Pin It on Pinterest

Share This