Artificial intelligence brings surgeons into the future

Orthopaedists must be able to quickly identify high-risk patients and evaluate surgical risks in relation to the benefits while considering the patient’s individual background and condition. Wrong decisions about treatments incur heavy costs, needlessly worry patients and make their lives more complicated.

Even experienced orthopaedic specialists find it difficult to accurately evaluate a patient’s surgical risks, especially if they have other illnesses. Risk factors are rather widely known – obesity is a major one, for example, but what if the patient uses a certain medication, suffers from a systemic illness or has certain blood values?

Taking the above into account, a strategy was chosen: combine the clinical knowledge of Coxa and the data science expertise of Solita with the aim of creating artificial intelligence models based on the surgical data that could predict surgical risks.

Over time, surgical data had been collected into several systems and databases, so the natural solution was to build a dedicated data lake to store all of the data and allow better processing. The preparation of the data required special care to ensure uniformity, quality and future utility.

Then we could proceed to the most interesting stage: Coxa’s research director and Solita data scientists began to research studies to find variables linked to risks in joint replacement surgery. They also experimented with different AI algorithms to develop risk prediction models. The options were neural network solutions and various machine learning models.

The machine learning models were built to keep the connections between determining variables visible. The number of input variables was also kept moderate to make the tool quick and practical to use, and to allow the impact of individual risk factors to be illustrated.

Taught and tested on separate data sets, dedicated prediction models were developed for each risk. National and international material was also found to independently validate the models. Furthermore, a peer review of the method and results was requested from a leading specialist, Professor Jonas Ranstam, who is an award-winning statistician at Lund University in Sweden, and he was impressed with the outcome.

Now in full working order, the prediction models were ready for clinical trials, so we developed a clear and handy browser-based user interface for orthopaedists. The tool was named Oravizio.

In Oravizio, the orthopaedic specialist needs to input just seven to fifteen patient variables, such as date of birth, height, weight, gender, joint to be operated, medication and laboratory results. At Coxa, Oravizio receives the patient information automatically from the data lake.

The Oravizio tool then combines this risk-related patient information and evaluates the patient’s personal surgical risks based on a massive data set. This helps identify patients who should not be operated on or who should improve their health before an operation.

• Risk of infection within one year from surgery: The probability that the operated joint becomes infected and requires a revision surgery as a result.
• Risk of revision within two years from surgery: The probability that a new operation is needed for any reason.
• Risk of death within two years from surgery: The probability that the patient will die within two years from surgery for any reason. Joint replacement surgeries are often performed on the elderly, thus the topic is highly important, especially with high identified risks.

Surgical risks are difficult for patients to understand and hard facts are not easy to provide.

Oravizio displays the potential risks to patients in a way that is clear and accessible. It also offers easy visualisation of different factors and their impact on risks, which can motivate lifestyle changes, for example.

The first CE-marked medical software for patient-specific joint replacement surgery risk evaluation

Because Oravizio is a tool that affects patient care, the software is governed by medical device regulations. This is why we decided to implement the methods of the ISO 13485 standard from early on in development. Clinical trials were an important part of development, and Coxa’s orthopaedic specialists used Oravizio during their surgery hours for about a year.

The results of the clinical trials were positive, so we applied for a medical device CE marking for the tool. The final product was the commercial Oravizio, the world’s first CE-marked medical software for patient-specific joint replacement surgery risk evaluation.

The Oravizio risk evaluation tool is unique in the world and brings surgeons into the future. Oravizio has attracted attention both in Finland and abroad. Pilots are currently underway in a number of hospitals in Finland and the Netherlands.