Doctor Roberto Salgado
"I have been actively involved in the integration of artificial intelligence (AI) in pathology for about 5 to 7 years, particularly focusing on oncology. My initial experience began with measuring biomarkers like tumor-infiltrating lymphocytes (TILs), which are crucial in cancer prognosis. This journey has highlighted the significant time and effort required to achieve clinical acceptance of new technologies. For instance, it took us nearly a decade to establish the routine use of TILs in clinical settings, illustrating the complexities involved in convincing healthcare providers, laboratories, and regulatory bodies to embrace new biomarkers.
In Belgium, the reimbursement process is particularly stringent. When a new AI tool is submitted for reimbursement consideration, the government delegates the assessment to an experienced entity called Sciensano. They evaluate the tool's performance, reproducibility, and clinical validity. However, if there are discrepancies between tools measuring the same biomarker—like the HER2 test—it raises concerns about treatment consistency. For example, if one laboratory's AI tool indicates a positive result for HER2, while another's shows a negative result, it could lead to either overtreatment or missed treatment opportunities for patients. This uncertainty makes the government reluctant to allocate funds for such technologies.
To navigate these challenges, it is important to foster collaboration between AI companies and pathology laboratories. One promising approach is to establish standardized guidelines and collaborative studies that evaluate multiple AI tools simultaneously. An example of such an initiative is the ongoing collaboration with the "Friends of Cancer Research" group in the United States, which aims to compare various AI tools measuring HER2 across different laboratories. The involvement of regulatory bodies like the FDA in these studies is vital, as it lends credibility to the findings. If companies like Owkin participate in these comparative studies, it may strengthen their case for reimbursement by demonstrating the reliability of their tools in real-world settings.
Moreover, I believe that investing in the digitalization of pathology laboratories is essential for market access. Currently, only a few laboratories in Belgium are fully digitalized, which limits the adoption of AI tools. If AI companies were to invest in providing digitalization solutions—such as scanners and software—this could facilitate the integration of AI tools into routine practice. For instance, Roche successfully collaborated with us in the past by funding genomic testing initiatives, which helped gather critical data for future reimbursement discussions. This model can be replicated for AI tools, where initial investments by companies could lead to more extensive adoption and, ultimately, reimbursement.
In conclusion, while there are significant barriers to the adoption and reimbursement of AI tools in pathology, I remain optimistic. The ongoing collaboration between academia, industry, and regulatory bodies is crucial to overcome these challenges. By focusing on rigorous validation, standardization, and strategic investments, we can pave the way for AI tools to transform patient care in oncology and beyond. I firmly believe that with the right approach, we can ultimately achieve a future where AI becomes an integral part of the diagnostic landscape, enhancing the accuracy and efficiency of patient care."
