The artificial intelligence (AI) revolution has created a remarkable opportunity to transform medical diagnostics by significantly enhancing the accuracy, speed, and efficiency of diagnosing patients.
AI algorithms analyze medical images, such as X-rays, MRIs, ultrasounds, CT scans, and DXA ( dual energy X-ray absorptiometry) scans. AI can process vast amounts of patient data, including 2D/3D imaging, bio-signals, electronic health records, as well as demographic information, medical history, and laboratory test results. These same algorithms can swiftly identify diverse patterns and anomalies, and possess the capability to interrogate extensive datasets, pinpoint biomarkers that clinicians might miss, and augment the precision of diagnosis and treatment.
While this application of data analysis is phenomenal in and of itself, the ability to take a holistic view of a patient from multimodal sources can help diagnosticians to find root causes of disease, treat patients more accurately, and even save lives.
Creating a market
The market for AI in diagnostics is poised for rapid growth through the end of the decade , buoyed by the emergence of startups and increased funding, with key players focusing on innovative product development strategies using AI-backed solutions in diagnostics.
Companies working to integrate AI into their diagnostics products include GE Healthcare, Philips Healthcare, Aidoc, and Siemens Healthineers . However, there are significant hurdles manufacturers must jump before products can make their presence felt on the market.
One of the primary barriers is the disagreement in data protection and the lack of compatibility with ethical aspects. The quality of training data is another significant hurdle, while the knowledge and trust of physicians in AI-supported systems can also pose a challenge. Navigating regulatory barriers presents another obstacle to manufacturers wishing to enter the diagnostics market. Moreover, there can be misaligned incentives within the healthcare system for these manufacturers that may not support the adoption of AI (Teodoridis, 2022).
AI tools today
AI-driven diagnostic tools are revolutionizing the accurate interpretation of medical images. These tools, powered by advanced machine learning algorithms, have garnered widespread acclaim, with numerous FDA approvals, particularly in radiology.
AI can empower healthcare professionals to devise more tailored and efficient treatment strategies, elevating the healthcare journey for patients and enhancing overall patient satisfaction. Beyond supporting patient-facing decisions, automation systems have the potential to find efficiencies in workflows, optimize resource allocation and save costs for healthcare systems.
As the data that AI generates grows in complexity, regulators and policymakers are raising concerns about potential bias, transparency and accountability. In response, researchers and practitioners have been actively developing and refining explainable AI (XAI) techniques to create a more transparent offering.
“Quantum AI is still in its early stages of development, and many technical challenges remain to be overcome before it can be widely deployed.”
AI-powered Clinical Decision Support Systems (CDSSs) offer real-time assistance for informed patient care decisions and will provide healthcare professionals with evidence-based recommendations and treatment guidelines tailored to individual patient characteristics. They could help clinicians to navigate complex medical scenarios, optimize treatment strategies, and improve patient outcomes.
However, as AI applications become more sophisticated, there is documented concern regarding ethics and liability (Jones, 2023).
Not only will support systems mature, but the use of AI in diagnostics will follow general industry trends such as personalized medicine, decentralized care, and a growing competitive marketplace.
The market will follow innovation
AI-powered diagnostic tools will increasingly leverage predictive analytics to forecast disease progression and identify individuals at risk of developing certain conditions. By using increasingly accessible longitudinal patient data to identify subtle patterns and trends, AI systems can help healthcare providers intervene proactively to prevent disease progression or complications.
The future of AI in medical diagnostics includes the development of portable and point-of-care diagnostic devices that leverage AI algorithms for real-time analysis. These devices could enable rapid and accurate diagnosis at the bedside, in remote locations, or even at home, thereby empowering patients and healthcare providers with timely information for decision-making.
New players will enter the market, with big tech companies such as OpenAI’s DeepQA and Google’s DeepMind developing general AI (GAI) applications for medical diagnostics, although it is still unclear if these will enter medical markets. As a myriad of companies and institutions begin to create AI-powered tools, this brings with it the necessity for interoperability standards and protocols to guarantee seamless collaboration and effectiveness among these tools.
The AI in Diagnostics pipeline
| Manufacturer | Device | Description |
| Philips Healthcare | Ultrasound Imaging System | Philips has submitted a patent for an ultrasound system titled “Ultrasound System with an Artificial Neural Network for Guided Liver Imaging.” The system is designed to enhance liver imaging by employing an artificial neural network (ANN) to precisely quantify the hepatic-renal ratio. The patent was published in June 2021, and the status is pending as of now. (Teodoridis, 2022) |
| Siemens Healthineers | Diagnostic Imaging | Siemens Healthineers submitted a patent for an AI-based product titled “AI driven longitudinal liver focal lesion analysis.” This method involves analyzing liver lesions using AI which includes the steps: curating masks from initial images, propagating regions of interest, performing assessments across images, selecting the best mask, and characterizing lesions across sequences and time points. The patent was published in July 2023 and the status is pending as of now. (Jones, 2023) |
| FUJIFILM | DiagnosticImaging | FUJIFILM has submitted a patent for a medical image processing device, a liver segment division method, and a program that is capable of dividing a liver into segments in a medical image that could be a CT scan, MRI image, ultrasound image, or PET scan image. The patent was first applied by FUJIFILM in July 2022 and published in March 2023. The status of the patent is pending as of now. (Clarivate, Derwent) |
Source: public records, Clarivate Cortellis *Patent applications do not establish that the invention is in compliance with applicable laws and regulations nor that the use of AI in that invention complies with the same.
Quantum AI is still in its early stages of development, however, and many technical challenges remain to be overcome before it can be widely deployed. Researchers are actively exploring various approaches to integrate quantum computing principles into AI algorithms and applications, with the goal of unlocking new capabilities and advancing the field of artificial intelligence. We cannot underestimate the growth of quantum computing to address challenges that are beyond the capabilities of classical computing. Quantum AI may enable more efficient optimization algorithms, faster machine learning model training, and improved pattern recognition in large datasets.
As with all artificial intelligence applications, this is just the beginning of the road less travelled. The future of AI in medical diagnostics is characterized by innovation, growth, and a commitment to improving patient care. By harnessing the power of AI technology, healthcare providers can unlock new insights, accelerate diagnostic workflows, and ultimately, save lives.
Read our newly-published report on the evolving regulatory landscape around in vitro diagnostics here. To learn more about how Clarivate helps medical device and diagnostics manufacturers deliver innovative products and solutions to patients, please visit us here.
References:
- Goldfarb, A., Teodoridis, F. (2022) Why is AI adoption in health care lagging?
- Jones, C. et al. (2023) Artificial intelligence and clinical decision support: clinicians’ perspectives on trust, trustworthiness, and liability. Medical Law Review
