Affidea

Dr Alessandro Roncacci

AI is said to bring incredible benefits to the clinical workflow. But what happens when smart technology, data, and humans join forces and work together seamlessly? Demonstrating the value of AI is a challenge in the real world. We are now witnessing the shift from whether AI can bring benefits in the clinical workflow and outcomes and how effectively we can evaluate the AI algorithm and implement it in a multi-country environment to ensure that it delivers the promised value. This is a question which remains open, and it depends on how mature the AI algorithm is and how structured and standardised the healthcare provider is.

At Affidea, we have an ideal environment in terms of geographical presence, multinational clinical expertise, best technology across 15 different healthcare markets, and a team of sub-specialty experts. 

Over the past two years, we have been involved in several pilots to assess the clinical, operational use, and commercial opportunities associated with AI-enabled technologies in Radiology. Our aim was to verify how some AI solutions could qualitatively support the reporting activity of our radiologists in specific disciplinary areas or how they can contribute to increase the safety of our patients who undergo diagnostic tests or to optimize the operational processes that govern our daily workflows. In all cases, we have expanded our experience and confirmed that artificial intelligence capabilities are developing every year, and technology can certainly ease the burden of overburdened resources without replacing the unique skills of doctors and practitioners of the healthcare sector. The smart integration of AI can improve the lives of everyone that interacts with i patients, medical personnel, and operations overall.

We have created a nine-stage framework for the evaluation and commercialization of AI solutions, including:

• selection criteria for AI solutions, countries, and centres;  

• legal review including medical device class and data protection impact assessment;

• technical architecture review including digital infrastructure and AI solution integration;

• training of healthcare professionals;

• clinical and technical assessment;

• identification of key performance indicators; 

• workflow redesigning to assess the benefits of the AI solution;

• commercialization process; 

• preparation of stakeholders’ information and communication strategy.

"Technology can certainly ease the burden of overburdened resources without replacing the unique skills of doctors and practitioners of the healthcare sector".

Using this framework, we clinically validated SubtlePET, a software that uses an AI algorithm to improve the quality of the PET/CT images that have been acquired with reduced dose or reduced scan time. In order to validate the algorithm, a pilot was carried out at Affidea IRMET center in Torino, Italy, a UEMN/EBNM and EARL accredited PET-CT site, to assess whether SubtlePET could produce images of adequate diagnostic confidence while reducing the administered F18-FDG dose from the standard institutional dose, and whether the outcomes were reproducible between different PET scanner models. 

Despite the significant challenges of the Covid-19 pandemic, the Affidea IRMET team were able to produce outstanding clinical outcomes, achieving a 33 per cent reduction in FDG dose without compromising image quality. The achievement was tremendous—a lower dose for patients that undergo PET-CT exams means greater safety with the same image quality for an accurate diagnosis, along with operational efficiencies for the medical center.  

The clinical team from Affidea IRMET center is now closing the second trial with the goal to achieve 50 per cent scan time reduction and with the ambition that in 2021 they will conclude the clinical validation and achieve scan time reduction in combination with FDG dose reduction. We trust that the scan time reduction will bring a great benefit to big oncological hospitals that have long waiting lists for PET-CT scans and need a trusted partner. In this way, we are best positioned to increase patients’ access to this examination when they need it most and offer our support to NHS and public hospitals.

Considering the successful results in the implementation of SubtlePET in our operationsat Affidea, we decided to roll out the AI implementation in other Affidea PET-CT centers from Italy, Romania, and Poland starting with January 2021.  

At Affidea, we have always embraced innovation and pioneered the latest technologies to improve clinical outcomes and patient safety. We will continue doing this, and we have big ambitions to achieve the same success in other clinical areas as breast, lung, and brain pathologies, where we are considering to integrate AI algorithms, following our structured framework for the evaluation and commercialization of AI solutions.