How Innovative Is China in the Robotics Industry?

Biomedical Engineer Integrates AI Techniques To Improve Diagnostic Medicine

Researchers at Rochester Institute of Technology developed new artificial intelligence techniques to extract and visualize information from standard-of-care biomedical data, providing a means for clinicians to better diagnose diseases and determine interventions. The new techniques could also improve image-guided therapies, including surgeries, and minimize invasive procedures because of these refined imaging details.

"The future of medicine is not necessarily about acquiring more data but rather having access to effective tools to make use of the data, and this is where biomedical computing plays a critical role," said Cristian Linte, professor of biomedical engineering in RIT's Kate Gleason College of Engineering. "Imaging accounts for the majority of biomedical data has transformed diagnostic and interventional medicine from a subjective, perceptual skill based on physicians' experience to an objective science driven by large-scale, heterogeneous data."

Computer-integrated diagnosis and therapy is an emerging field dedicated to improving disease detection and treatment. Linte and members of his research team, including Imaging science doctoral students Bipasha Kundu, Bidur Khanal, Zixin Yang, Nakul Poudel, and Richard Simon, detailed results of this work in several publications, including the April 2025 proceedings of SPIE Medical Imaging 2025.

Biomedical visualization has evolved from anatomical drawings to a standard tool to aid diagnosis, plan treatment options, and monitor therapy. Before biomedical data can be visualized, the raw biomedical imaging data needs to be processed. Integration of artificial intelligence (AI) into medical image analysis has led to significant advances, but several challenges still exist, Linte said. 

AI models rely on large amounts of expert-annotated data for training, which requires time and expertise of clinicians to curate data. User variability also poses a significant barrier for accurate AI algorithm development. Internal operations and relevance of test data acquired for training of AI models are also not well understood, making predictions difficult to explain.

 "Many physics-based biomedical models are hampered by their computational expense, which constitutes a major setback to clinical adoption, limiting their use as interactive simulation tools for therapy planning or monitoring," Linte said.  "AI techniques, on the other hand, can learn from large patient-specific datasets, so combining data science with physics-based models has the potential to yield more accurate and more computationally efficient simulations." 

Researchers in Linte's lab have effectively combined biomedical imaging, computing, modeling, and visualization for computer-integrated diagnosis and therapy. They contributed to the development and validation of robust AI computational imaging informatics tools to advance computer-integrated diagnosis and interventional data science by addressing a broad range of diseases, organ systems, and minimally invasive therapy applications.

 "We believe that effective utilization of biomedical informatics to develop versatile biomedical computing and visualization tools will lead to solutions that enable more accurate and timely disease diagnosis and less invasive therapies. These tools will help lay a foundation for advances in computer-aided diagnosis and therapy across a wide spectrum of diseases and organ systems that can impact a larger patient population," said Linte, who has a background in mechanical and biomedical engineering as well as imagining science. He teaches in RIT's engineering college as well as the Chester F. Carlson Center, specifically in the areas of biomechanics and biomedical thermo-fluids and conducts research at the intersection of biomedical imaging, computing, and visualization.

Research in Linte's Biomedical Imaging, Modeling, Visualization and Image-guided Navigation Laboratory is supported by grants from the National Institutes of Health (NIH) and the National Science Foundation. Its research focus remains on biomedical artificial intelligence tools for diagnostic and intervention data science. Most recently, he was awarded nearly $2.4 million by the NIH for a five-year competitive renewal of the research grant on Biomedical Computing and Visualization Tools for Computer-integrated Diagnostic and Therapeutic Data Science to support innovation, training, and mentorship of graduate and undergraduate students in the lab, many of whom have gone on to serve in prestigious national labs, hospitals, and research organizations.

"Mentoring and training high caliber students who will join tomorrow's biomedical and academic workforce constitutes by far the greatest impact of our careers as academics and scientists and we're thrilled to see them succeed," said Linte.

How AI Is Revolutionizing Rare Disease Diagnosis

AI tools are shortening diagnostic timelines and providing timely, accurate answers for rare disease patients

The seven-year medical odyssey that once defined rare disease diagnosis is being rewritten by algorithms capable of identifying medical zebras in the time it takes to refresh your social media feed. Artificial intelligence tools are slashing diagnostic timelines from years to seconds, potentially saving lives and eliminating the painful "diagnostic odyssey" that rare disease patients have endured for generations.

These AI diagnostic systems are rapidly transforming from theoretical possibilities to practical clinical tools, analyzing everything from facial features to genetic sequences with a speed and accuracy that can surpass even experienced specialists. For the estimated 400 million people worldwide living with rare diseases, these technologies offer something precious that many have never experienced—the simple dignity of a timely, accurate diagnosis.

Let's explore how these remarkable AI tools are revolutionizing rare disease detection and why they might represent the most significant advancement in diagnostic medicine since the invention of the microscope.

The pattern recognition breakthrough

Rare diseases present a perfect paradox for human diagnosticians. Their unfamiliarity makes them difficult to recognize, yet many leave subtle patterns of symptoms, lab results, and physical manifestations that follow consistent, identifiable rules. This combination creates the perfect problem for artificial intelligence to solve.

AI systems excel at detecting patterns too subtle or complex for human recognition, particularly when trained on massive datasets encompassing thousands of rare conditions. While even the most experienced specialist might encounter only a handful of cases of a particular rare disease in their entire career, AI tools can analyze millions of patient records, scientific literature, and clinical images to identify the telltale signatures of even the most obscure conditions.

This pattern recognition capability works across multiple data types. Facial analysis algorithms can identify distinctive facial features associated with hundreds of genetic syndromes from a single photograph. Natural language processing systems can scan years of clinical notes to spot symptom combinations that point to specific rare diseases. Image analysis tools can detect subtle anomalies in radiology scans that might escape human notice.

What makes this approach particularly powerful is how these systems improve over time. Each new case expands the AI's knowledge base, making future diagnoses more accurate and comprehensive. This continuous learning creates a virtuous cycle where diagnosis becomes increasingly precise as the technology is more widely adopted and exposed to more diverse patient data.

The genetic analysis acceleration

Perhaps nowhere has AI made a more dramatic impact than in the interpretation of genetic data. While genetic sequencing costs have plummeted, the expertise needed to analyze these results remains scarce. A single patient's genome contains roughly 3 billion base pairs, making manual analysis practically impossible for complex cases.

AI systems can now scan genetic sequences and identify disease-causing mutations in minutes rather than the weeks or months traditional analysis might require. These tools compare patient genetic data against databases of known disease-causing variants while also evaluating previously unclassified genetic changes for their potential to disrupt normal biological function.

The most advanced AI genetic analysis tools go beyond simple pattern matching to predict how specific genetic variants might affect protein structure and function. By simulating these molecular interactions, AI can help determine whether an unfamiliar genetic change is likely to cause disease, expanding diagnostic capabilities beyond the limitations of existing medical literature.

This capability is particularly valuable for ultra-rare conditions where limited case data exists. Many patients with extremely unusual genetic disorders have spent years without diagnosis because their specific genetic variant hasn't been previously documented. AI-powered predictive analysis can provide crucial insights even for these one-in-a-million cases, suggesting likely diagnoses where no human reference point exists.

The symptom timeline detection

Many rare diseases evolve in distinctive patterns over time, with symptoms appearing in specific sequences or combinations that serve as diagnostic signatures. Traditional medical approaches often struggle to connect these temporal dots, especially when symptoms develop over years and span multiple medical specialties.

AI diagnostic systems excel at analyzing these longitudinal patterns, connecting seemingly unrelated symptoms separated by years into coherent disease timelines. By mining electronic health records for temporal relationships between symptoms, laboratory values, and treatments, these systems can identify disease trajectories that would be nearly impossible to detect through conventional medical approaches.

What makes this capability particularly valuable is how it works with incomplete or fragmented medical records. Even when patients have received care across multiple unconnected health systems, AI tools can piece together partial clinical narratives to suggest possible rare disease diagnoses that explain the full symptom constellation.

This temporal pattern recognition helps identify rare diseases at earlier stages, when treatment often proves most effective. By flagging subtle early manifestations that might otherwise be dismissed as unrelated to a patient's primary symptoms, AI systems can compress the diagnostic timeline from years to months or even weeks, potentially altering disease progression before irreversible damage occurs.

The literature-clinical connection

Medical knowledge about rare diseases exists in fragmentary form across millions of research papers, case reports, and clinical guidelines that no human physician could possibly master. AI systems bridge this knowledge gap by continuously scanning and synthesizing the entire body of rare disease literature and comparing it against individual patient presentations.

These AI literature review systems can identify relevant research about conditions that local physicians might never have encountered, essentially democratizing specialist-level knowledge. A doctor in a rural clinic can now tap into the same rare disease information that might previously have been known only to a handful of researchers at specialized academic centers.

What makes this approach particularly powerful is how it handles the constantly evolving nature of medical knowledge. While human experts might take years to become aware of newly discovered conditions or treatment approaches buried in specialized journals, AI systems can incorporate this information immediately, ensuring patients benefit from the very latest scientific insights.

This literature-clinical connection helps solve one of rare disease diagnosis's greatest challenges—the "known unknown" problem where the correct diagnosis exists in medical literature but never gets connected to the patient's case. By systematically comparing patient data against the entire corpus of medical knowledge, AI dramatically reduces the chance that a recognizable rare disease will go undiagnosed simply because local providers aren't familiar with it.

The multisystem integration advantage

Rare diseases often affect multiple body systems simultaneously, creating symptom constellations that cross traditional medical specialties. This multisystem nature frequently contributes to diagnostic delays, as specialists focus on symptoms within their domain without recognizing the broader pattern.

AI diagnostic systems excel at this cross-specialty integration, simultaneously analyzing cardiological, neurological, dermatological, immunological, and other findings to identify conditions that might be missed when each system is evaluated in isolation. This holistic perspective helps detect diseases that manifest across multiple specialties but might appear non-specific when viewed through any single clinical lens.

The integration advantage extends to multiple data types as well. By simultaneously analyzing structured data like lab values, unstructured information from clinical notes, imaging studies, and even patient-reported symptoms from questionnaires, AI systems create comprehensive clinical pictures that more closely match how rare diseases actually manifest in real patients.

This multidimensional approach proves particularly valuable for conditions with variable presentations. While individual patients with the same rare disease might have different predominant symptoms, AI systems can recognize the underlying pattern across this variability, helping identify conditions even when they don't present in textbook fashion.

The real-world implementation challenges

Despite their transformative potential, AI rare disease diagnostic tools face significant implementation hurdles that have slowed their adoption in everyday clinical settings. Understanding these challenges helps set realistic expectations for how quickly these technologies will reach patients beyond academic medical centers.

Data privacy concerns represent a primary obstacle, as effective rare disease AI requires access to sensitive patient information. Ensuring these systems operate within regulatory frameworks while still accessing sufficient training data remains an ongoing challenge. This tension is particularly acute for pediatric rare diseases, where additional protections for minors' data must be navigated.

Integration with existing clinical workflows presents another hurdle. Even the most accurate AI system provides limited value if physicians find it cumbersome to use or if it generates results that can't be easily incorporated into clinical decision-making. Successful implementation requires thoughtful interface design and seamless embedding within electronic medical record systems that doctors already use.

Perhaps most challenging is the "black box" problem, where AI systems reach diagnostic conclusions through processes too complex for human comprehension. For physicians legally and ethically responsible for diagnostic decisions, this lack of explainability creates understandable hesitation. The most promising implementations address this by providing not just diagnostic suggestions but also transparent reasoning and confidence levels that help clinicians understand why the AI reached specific conclusions.

The patient perspective transformation

While technical capabilities rightly dominate discussions of AI diagnostics, their most profound impact may be on the lived experience of rare disease patients and their families. After generations of being told their symptoms were psychosomatic, exaggerated, or simply medically inexplicable, patients are finding validation and answers through these new diagnostic approaches.

The psychological toll of diagnostic uncertainty can be devastating. Patients often report that receiving an accurate diagnosis, even for a serious condition, brings immense relief after years of being dismissed or misdiagnosed. AI tools are dramatically shortening this period of uncertainty, allowing patients to more quickly access appropriate treatments, connect with relevant support communities, and make informed life decisions.

This diagnostic clarity particularly benefits patients from marginalized communities whose symptoms have historically been more likely to be dismissed by medical professionals. AI systems, when properly designed and trained on diverse datasets, can provide more objective evaluations that depend less on provider biases or communication styles, potentially helping address long-standing healthcare disparities.

Beyond individual diagnosis, these technologies are expanding our understanding of rare diseases themselves. By identifying previously unrecognized patients and detecting subtle phenotypic patterns, AI tools are refining condition definitions and revealing new disease subtypes. This expanded knowledge benefits both current and future patients by enabling more personalized treatment approaches based on specific disease mechanisms.

The rise of AI rare disease diagnostics represents one of those rare technological advances that delivers benefits across the entire healthcare ecosystem. Patients receive faster, more accurate diagnoses. Physicians gain powerful new tools to solve their most challenging cases. Researchers discover new disease insights and potential treatment targets. And healthcare systems potentially reduce costs by eliminating years of unnecessary tests and inappropriate treatments. In a medical field often defined by agonizing waits and disappointing dead ends, these AI tools offer something truly revolutionary—the gift of answers measured in seconds rather than years.

Lung Cancer Risk In Never-smokers Predicted By AI Tool 'Sybil'

ST. PAUL, Minn., May 19 (UPI) -- With lung cancer rates among non-smokers rising, especially young East Asian women, a new study released Monday is touting the promise of an artificial intelligence tool to "strongly" predict who's most at risk.

Lung cancer has long been associated with smoking. But even as overall rates steadily drop and smoking decreases around the world, a unique population of young East Asians are seeing a 2% annual increase in lung cancer cases -- even though half of them have never smoked.

The cause of this remains unknown, but suspicion is centered on genetic mutations developed during a person's lifetime rather than inherited, such as damage to a gene that codes for a protein known as EGFR, which prevents cells from growing too quickly.

This genetic damage is believed to be caused by environmental toxins including second-hand smoke and even fumes produced by high-temperature stir-fry cooking in rooms that lack proper ventilation.

Globally, more than 50% of women diagnosed with lung cancer are non-smokers, compared to 15% to 20% of men. Meanwhile, an estimated 57% of Asian-American women diagnosed with lung cancer have never smoked, compared to only about 15% of all other women, according to a recent University of California-San Francisco study.

Against this backdrop of rising cancer cases among seemingly low-risk women, the potential of AI to accurately predict who may be most suspectable to a surprise lung cancer diagnosis has generated considerable interest around the world.

In a paper presented Monday at the American Thoracic Society's medical conference in San Francisco, Dr. Yeon Wook Kim of the Seoul National University Bundang Hospital reported a new AI tool dubbed "Sybil" has proven to be accurate in identifying which "true low-risk individuals" are more likely to develop lung cancer -- all foretold from a single low-dose chest CT scan, or LDCT.

Sybil, named after the female seers of ancient Greek mythology, was developed in 2023 by researchers at the Massachusetts Institute of Technology's Abdul Latif Jameel Clinic for Machine Learning in Health, the Mass General Cancer Center and Chang Gung Memorial Hospital in Taiwan.

It was trained first by feeding it LDCT images largely absent of any signs of cancer, since early-stage lung cancer occupies only tiny portions of the lung and is invisible to the human eye. Then, researchers gave Sybil hundreds of scans with visible cancerous tumors.

In its first run, Sybil was able to deliver "C-indices" of up to 0.81 in predicted future occurrences of lung cancer from analyzing one LDCT. Models achieving predictive C-index scores of over 0.7 are considered "good" and those over 0.8 are "strong."

This week's Korean study validated those results. Kim and his colleagues evaluated 21,087 people ages 50 to 80 who underwent self-initiated LDCT screening between January 2009 and December 2021 in a tertiary hospital-affiliated screening center in South Korea. These subjects were followed up until June 2024.

Baseline LDCTs were analyzed with Sybil to calculate the risk of lung cancer diagnosis within one to six years. Analyses were performed for individuals with various smoking histories, ranging from more than 20 "pack-years" to never-smokers, who comprised 11,098 of the participants.

Among all participants, 257 (including 115 never-smokers) were diagnosed with lung cancer within six years from the baseline LDCT. Sybil achieved a C-index for lung cancer prediction at one year of 0.86 and 6 years of 0.74 for all the participants, while among never-smokers, one-year and six-year C-indices were 0.86 and 0.79, respectively.

Kim told UPI the results hold the promise of helping to regularize lung cancer screening in Asia, where those efforts are inconsistent and, due to differing demographics, sometimes are at a "disconnect" with international screening criteria.

"Asia bears the highest burden of lung cancer, accounting for over 60% of new cases and related deaths worldwide," he said in emailed comments. "A growing proportion of this burden is observed among individuals who have never smoked, particularly among women.

"In Korea, more than 85% of female lung cancer patients are non-smokers. As a result, increasing attention has been given to evaluating the effectiveness of lung cancer screening, or LCS, in traditionally low-risk populations in Asia."

Government-led programs and initiatives have expanded to include never-smokers into their LCS efforts, while other efforts varying from international guidelines due to their inclusion of such never-smokers have "gained traction in East Asian countries, including South Korea, Taiwan and China," Kim said.

AI tools like Sybil could be used to develop "personalized strategies" for patients who have already undergone LDCT screening, but have not yet had follow-ups, he added, while cautioning that further validation will be needed "to confirm the model's potential for clinical use.

"While the need for screening low-risk groups may be justified in certain settings, the lack of evidence from randomized trials limits the development of long-term LCS strategies for these populations."

Researchers, meanwhile, are "actively" working on expanding Sybil's uses into other personalized health applications, said Adam Yala, an assistant professor at the UCSF/UC-Berkeley Joint Program in Computational Precision Health and one of the AI model's developers.

"One, this is broadly applicable across many different types of cancers," he told UPI. "We've got processes ongoing for breast cancer, and we're also working on prostate and pancreas cancers.

"And there's also evidence that from CT scans you could predict sudden deaths from cardiovascular disease. This would provide early detection, giving you a better opportunity for early intervention to provide better outcomes. So it's not uniquely about cancer. ... There's a version of this for cardiovascular health, and there could be other areas of medicine, as well."

AI's potential to provide health benefits, Yala added, "is totally untapped. For instance, now we're only looking at a patient's CT scan once, but over time, you could look at multiple CTs. Mammograms, as well. There's a lot of data available there. It's a field at its infancy."

AI tools like Sybil have the potential to make screening much more efficient and personalized, said Dr. Jae Y. Kim, thoracic surgery chief at City of Hope, a U.S. Cancer research and treatment organization in Los Angeles County, Calif.

"Our current screening recommendations are for large populations, but we know that for a given group of people, it's nearly impossible to predict who will get cancer and who won't," he told UPI. "We have some algorithms that use very basic characteristics such as age, gender and smoking history to calculate a crude risk score for developing lung cancer. But these traditional risk models fail populations like Asian women who have never smoked because it underestimates their lung cancer risk."

If lung cancer screening is expanded to include non-smokers, it would mean "a lot more people getting a lot more CT scans," but with AI's ability to give a much more accurate prediction of cancer risk, "the potential benefit is that it can identify people who might be at higher risk who should perhaps get screened more frequently."

Meanwhile, others who are at lower risk and may need to get screened less frequently or not at all.

"This could prevent unnecessary testing for a lot of patients and result in cost savings for our health system," Kim said.

---