Artificial intelligence (AI) has surged in popularity, becoming a favored conversation within business and media. The infrastructure speed, accessibility, and scale has allowed companies to develop more ambitious and accessible algorithms that can be applied to a greater range of datasets, improving current programs, or creating entirely new ones. While enhancing capabilities across multiple sectors, AI continues to bring forth groundbreaking advancements in healthcare, revolutionizing the lives of patients, families, and clinicians. Evidencing its great promise in healthcare, AI is currently being used to advance our ability to detect and diagnose several diseases and neurodegenerative disorders.
AI digital diagnostics is a growing field that uses data analytics and machine learning (ML) to develop innovative and non-invasive solutions for diagnosing patients quickly, accurately, and, in many cases, remotely. AI is transforming healthcare diagnostics by deriving new and significant insights from evolving databases. Two main applications of AI in diagnostics are medical imaging analysis and remote diagnostic monitoring, followed by diagnostic decision support systems and point-of-care testing. AI digital diagnostics represents an innovative and practical space that aims to standardize and surpass the abilities of human diagnoses by avoiding error and subjectivity and relying on ever-changing data to stimulate growth and accuracy.
Medical Imaging Analysis
AI digital diagnostics are being used to improve the speed, accuracy, and detection capabilities of medical images, such as MRIs, X-rays, and Ultrasounds. Most AI algorithms currently focus on rapid analysis and interpretation of images, identifying subtle patterns and abnormalities that go beyond human detection. They support doctors in the diagnosis of cancerous tissue, brain lesions, and neurodegenerative disorders. Furthermore, advancements in medical imaging analysis have revolutionized the way healthcare professionals prioritize patient triage, ensuring that the most critical cases receive appropriate care.
AI algorithms are trained and improved by continuously collecting large, diverse datasets of labeled medical images and/or patient data, which serve as the foundation for training algorithms. Feature extraction techniques are then used to identify patterns in the medical images, which function as model inputs, and work to finetune the model’s predictions. The model is tested for sensitivity and specificity and is deployed for real-world use in digital diagnostics. Workflow integration of medical imaging analysis software can take place in a few ways: the AI can have an API that communicates with the pre-existing radiology system, the AI can be made compatible with data formats like DICOM for medical images, or the AI can be integrated into the radiology workflow through PACS or RIS, two systems commonly used in radiology departments. Through this integration, doctors can avoid any inconsistency in detection, creating an objective device that mitigates human error and variability.
Beginning in 2010 with the introduction of deep learning, medical imaging analysis for diagnostics has witnessed revolutionary growth. More than $2.2B has been invested in new startups, with the investment since 2017 being 200% higher than the total since 2010. This uptick is likely influenced by the increased demand for diagnostic procedures. Chronic diseases are becoming more common and healthcare systems are overburdened, making AI a practical and useful tool. AI in medical imaging analysis has great potential for growth, with an expected CAGR of 33% between 2023 and 2030. However, the speed at which medical imaging companies have entered the market suggests that this space is highly competitive, with the peak of company influx predicted to be realized between 2025-2027.
The anticipated growth of AI in medical imaging analysis, coupled with FDA clearance of over 500 AI-medical imaging software solutions, suggests rapid advancement and saturation within the field. Thus, emerging companies have explored various avenues for differentiation by applying their unique algorithms to multiple modalities, including MRI scans, Ultrasounds, and X-rays. For instance, the medical imaging company EnvisionIt Deep AI uses a proprietary platform to cover multiple radiology sectors, and with that, hundreds of conditions. Its platform is applicable and teachable, making it a seamless integration into hospital systems and addressing the challenge of clinician skepticism. Moreover, some well-established companies, like HeartFlow, target a single indication, like cardiovascular or neurodegenerative diseases. By leveraging multiple modalities to best specialize in the diagnosis and treatment of a given indication, these companies seek to become overarching platforms for specific patient populations.
The true success of AI for digital diagnostics hinges on the algorithm’s data inputs, detection accuracy, and applicability to various diseases. Emerging companies like Imeka, Darmiyan, and Care Health are medical imaging companies dedicated to improving the speed and precision of brain imaging, either for the purpose of detecting changes in brain tissue or identifying intracerebral hemorrhaging. Digital diagnostics oriented toward the detection of neurodegeneration are particularly exciting because, despite significant progress, these diseases remain challenging and complex. More research is needed to discover effective treatments and, ultimately, potential cures. Other companies, such as Prevonotics and RevealDX, direct their efforts toward non-invasive cancer detection and diagnosis, prioritizing early detection to increase treatment outcomes. Medical imaging analysis is an emerging modality with great potential to improve the efficiency of the healthcare system and empower patients to be proactive in their health journey.
Remote Diagnostic Monitoring
Remote diagnostic monitoring (RDM) is another area for disruption with AI. RDM enables healthcare workers to monitor and diagnose a patient remotely by combining technology and patient data. Increasingly, RDM platforms leverage AI algorithms to analyze and interpret large volumes of patient data collected through a remote device, whether that be a smartphone-image, a wearable device, or a testing kit. The remote diagnostic monitoring market is valued at $4.4B in 2022, with an expected CAGR of 18.2%. Notably, many of the advancements within AI-RDM have targeted pediatrics, with the hope of reducing unnecessary urgent care visits for minor illnesses and conditions among children. A quick diagnosis and treatment plan is transformative in today’s age.
The AI-RDM space can be segmented based on the type of device being leveraged. Many AI-RDM companies develop platforms that use smartphone imaging to analyze a part of a patient’s body and determine a condition and its severity within minutes. Other companies use wearable devices, like bracelets and heart monitors, to detect abnormalities in heart rhythms and breathing over a longer course of time. Across these two modalities, AI-RDM systems allow patients to avoid unnecessary hospital visits and allow providers to immediately develop treatment plans, thereby reducing overall time and costs while improving outcomes.
Smartphone-based systems have emerged as the newest wave in RDM. Numerous advancements have spurred the uptick in such systems, including increased smartphone accessibility, improvements in camera optics, and the adoption of telemedicine and remote consultations. Smartphone-based RDM allows patients to circumvent a visit to their provider or urgent care center by diagnosing specific conditions from images taken in the comfort of their homes. Given the relative nascency of the space, companies have primarily focused on single indications. For instance, WavelyDx diagnoses ear infections, while Momentum Health diagnoses scoliosis. While this diagnostic specificity has likely allowed companies to develop thorough algorithms, it has resulted in a landscape consisting of siloed players. The few smartphone-based RDM companies developing a portfolio of diagnostic capabilities for a more-generalized patient population or specialty, like DermaDetect for dermatology, are likely to have a greater impact. With the majority of startups operating within specific conditions, it has yet to be determined whether companies will expand their diagnostic capabilities internally or through acquisitions. No matter the means, companies aggregating multiple smartphone-based diagnostic capabilities are poised to serve as invaluable access points for patients who are otherwise fatigued from navigating across platforms.
Wearable devices serve as valuable tools for patient diagnosis and monitoring. Typically in the form of a bracelet, these devices track bodily functions to identify abnormal conditions. For instance, the RDM company Alva Health uses a bracelet to detect signs of a stroke. Despite addressing critical issues like stroke detection, many of these bracelet-based RDM companies leverage common data inputs, like heart rate and breath. With the rise of digital watches in the mainstream, bracelet-based RDM companies’ risk being leapfrogged by big players, like Apple, using their already-popular devices to generate similar diagnostic outputs. However, some RDM companies offer a unique advantage by developing AI-driven devices for behaviors that would otherwise need to be observed by a doctor. For instance, the RDM company ClearSky Medical Diagnostics develops hands-on medical devices that are specific to tracking neurodegenerative disorders. This company has developed a glove that tracks the tremors experienced by an individual with Parkinson’s Disease to determine their medication needs. Through novel integrations of devices and AI software, these companies differentiate themselves within the space.
As a well-established space, RDM requires unique innovation for companies to stay relevant and successful. Moving away from wearable device monitoring and towards integrating AI algorithms into smartphone devices for diagnosis of common conditions is an advantageous way to stay ahead in the market. However, to be differentiated within the smartphone-based RDM space, companies need to be highly accurate and cover multiple indications. Within pediatrics, for instance, a platform leveraging smartphone images to diagnose multiple conditions with immediate results would be advantageous. In our highly tech-enabled society, embracing a direction that allows individuals to be diagnosed from the comfort of their home, avoiding wait times and reducing costs, is a wise move for startups and investors.
AI digital diagnostics have been leveraged for point-of-care testing (POTC) to guide patient triage and identify acute conditions. POCT is often conducted quickly and can be less reliable, making objective AI algorithms highly beneficial assurance tools to enhance the detection and interpretation of tests. AI will transform POCT, making it cheaper, quicker, and quality-assured. The global POCT market size was $43.93B in 2022 and is projected to grow at a CAGR of 7.9%. Notably, however, this is an overarching market size in POTC, and not specific to AI-POCT.
Companies like Biocogniv and Pons Digital Health are at the forefront of utilizing AI-algorithms in POTC to assess a patient’s severity and determine their need for surgery, hospitalization, or immediate treatment. Biocogniv is a digital diagnostics company that tackles time-critical conditions using AI and patient data. Their software platform is connected to multiple hospital systems and developed within a cloud-based digital diagnostic tool that enables hospital providers to identify patients with an acute condition using only routine laboratory tests and AI. Pons, on the other hand, provides a mobile ultrasound linked to an algorithm so that anyone, specifically nurses, can avoid waiting for a doctor or lab result to determine patient severity.
While POTC is growing at a slower rate than medical imaging analysis and RDM, the introduction of AI into routine POTC is likely to catalyze market growth. By enhancing the efficiency of routine tests, like an ECG for instance, hospitals can swiftly determine the urgency of a condition and optimize the organization of their emergency room. AI-POCT systems that prioritize augmenting the capabilities of existing tests are likely to succeed.
AI digital diagnostics will continue to transform the detection of diseases and illnesses, revolutionizing the speed and accuracy of radiology images and remote medical devices. In a world where hospitals are facing a severe shortage of healthcare professionals and a simultaneous increase in chronic diseases, AI diagnostic support tools will be crucial to progressing healthcare capabilities. Furthermore, by reducing physician oversight and leveraging relatively accessible devices, digital diagnostics promise to democratize healthcare, particularly for underserved communities. As the digital diagnostics field continues to advance with sophisticated algorithms and unique modalities, companies oriented towards expanding access to diagnostic care will have a far-reaching impact on the healthcare system. Although a wide and diverse field, companies carving out niches—like neurological medical imaging analysis—or alternatively aggregating capabilities—like pediatric RDM—within digital diagnostics are poised to succeed.