How to Improve Cancer Diagnostics With Technology

AI just ran its own multidisciplinary tumor board. And nailed the diagnosis + treatment. This was a full-stack oncology reasoning engine—pulling from imaging, pathology, genomics, guidelines, and literature in real time. A new paper in Nature Cancer describes how researchers built a GPT-4-powered multitool agent that: • Interprets CT & MRI scans with MedSAM • Identifies KRAS, BRAF, MSI status from histology • Calculates tumor growth over time • Searches PubMed + OncoKB • And synthesizes everything into a cited, evidence-based treatment plan In short: it acts like a multidisciplinary team. Results : • Accuracy jumped from 30% (GPT-4 alone) to 87% • Correct treatment plans in 91% of complex cases • Every conclusion backed by a verifiable citation This is bigger than oncology. Any field that relies on multi-modal data and cross-domain reasoning—like my field of GI ( GI + Mental Health+ Nutrition + Excercise ) could benefit from this collaborative AI architecture. Despite the visual, it doesn’t replace the human team—it augments it. Providers still decide. But now, they do it faster, with more context, and less cognitive fatigue. #AI #HealthcareonLinkedin #Healthcare #Cancer

I'm excited to announce the publication of our team's latest work in npj Precision Medicine! We've developed MetaCH, a machine learning framework that improves the interpretation of liquid biopsies in cancer care. The promise of circulating tumor DNA (ctDNA) lies in its potential for disease monitoring and early diagnosis, offering a less invasive approach than traditional tumor biopsies. However, an obstacle to unlocking this potential is distinguishing true tumor-derived mutations from those arising from clonal hematopoiesis (CH), or age-related mutations in blood cells. MetaCH tackles this challenge by accurately classifying CH variants using only cell-free DNA from plasma samples, bypassing the need for costly and time-consuming matched white blood cell sequencing. MetaCH achieves this through a unique three-stage process: 🧬 The Mutational Enrichment Toolkit (METk) generates context-aware representations of mutations by integrating sequence context, gene information, and cancer type, capturing a more comprehensive picture of the mutational landscape. 🤖🤖🤖 Base classifiers trained on both large-scale public cancer and blood genomic datasets and a smaller, more detailed matched cfDNA dataset allow us to leverage the breadth of general cancer knowledge alongside the specificity of matched samples to score the CH-likelihood of variants. 🎯 A meta-classifier integrates the scores from the base classifiers, providing a final prediction of variant origin (tumor vs. CH). 🚀 MetaCH surpasses current classification methods across multiple types of cancer datasets to improve the accuracy of liquid biopsy-based cancer diagnostics and monitoring. ➡️ Learn more about MetaCH and its potential to transform cancer diagnostics: https://lnkd.in/eMhxhwNt Thanks to all co-authors! Gustavo Arango, Marzieh Haghighi, Gerald Sun, Elizabeth Choe, Aleksandra Markovets, J.Carl Barrett, Zhongwu Lai #PrecisionMedicine #AI #MachineLearning #CancerResearch #LiquidBiopsy #AstraZeneca #Oncology #ctDNA

🎗️ Transforming Cancer Care with AI: The Game-Changing Power of CHIEF 🎗️ Harvard Medical School introduced CHIEF (Clinical Histopathology Imaging Evaluation Foundation), an advanced AI model set to revolutionize cancer diagnosis, treatment guidance, and survival predictions across 19 cancer types. This versatile tool, detailed in Nature, opens new possibilities for patient care and personalized treatment. CHIEF was initially trained on 15 million unlabeled images and then further refined on 60,000 whole-slide images, allowing it to interpret both specific sections and broader image context for a holistic understanding. 🔬 Broad Diagnostic Capabilities Across Multiple Cancers Trained for multiple tasks, CHIEF detects cancer cells, predicts outcomes, and analyzes molecular profiles. Achieving 94% accuracy, it surpasses existing models, proving highly adaptable in varied clinical settings. 🧬 Advanced Molecular Profiling CHIEF efficiently fills gaps in traditional DNA sequencing by analyzing cellular patterns to predict genetic mutations. It achieved over 70% accuracy in identifying 54 key cancer genes, making treatment personalization quicker and more accessible worldwide. 📉 Predicting Patient Survival with Accuracy CHIEF forecasts survival with precision, distinguishing patients with high versus low survival rates based on histopathology. It outperformed other models by 8-10%, aiding early identification of patients for targeted treatments. 📊 Novel Insights into Tumor Behavior Beyond diagnostics, CHIEF uncovers new insights, identifying cellular patterns linked to survival, such as higher immune cell presence, potentially guiding future biomarker development for cancer aggressiveness. 🧩 Future Steps for Enhancing CHIEF Plans include additional training on rare diseases, expanding molecular data, and refining its ability to predict outcomes for emerging therapies. Summary: CHIEF exemplifies AI’s transformative potential in cancer care, making diagnostics faster, more accurate, and tailored. This powerful tool offers hope for patients and oncologists alike by advancing personalized cancer treatment. #AIinHealthcare #HarvardMedical #CancerDiagnosis

Pathogenomics is an emerging approach to cancer diagnosis that integrates genomic data, morphological information from histopathological imaging, and codified clinical data to better capture tumor heterogeneity while enabling the discovery of new multimodal cancer biomarkers. In a new review written by @Xiaobing Feng, the authors analyze recent pathogenomic studies that combine morphological information from histopathology and molecular information from genomic profiles to better quantify the tumor microenvironment and harness advanced machine learning algorithms for biomarker discovery. Pathogenomics for accurate diagnosis, treatment, prognosis of oncology a cutting edge overview. https://lnkd.in/e5UFAsaD Methods overview: The authors analyzed various approaches and techniques used in pathogenomics research. They explained the use of whole slide images (WSI) and computational pathology techniques to extract detailed pathological information, including nucleus shape, texture, global structure, and tumor-infiltrating lymphocytes patterns. The researchers described the application of advanced machine learning algorithms to automatically identify and quantitatively analyze important tissues and cells in images. They also discussed the use of deep learning models, such as deep residual learning (Resnet 18), to predict molecular features of tumors directly from H&E histology. The authors explained how different multimodal fusion approaches, including early fusion, late fusion, and hybrid fusion, detailing how these methods combine data from different modalities. They also described various interpretability techniques used in pathogenomics, categorizing them into ante-hoc and post-hoc explanatory methods. Results overview: The authors synthesized and discussed results from various studies in the field of pathogenomics, reporting  on studies that successfully correlated pathological morphology with large-scale genomic analysis. A notable successful example being the  prediction of genomic biomarkers like microsatellite instability status directly from H&E histology. They also described work showing that combining gene expression data with histopathological features to improve the accuracy of prognosis prediction in various cancer types. They highlighted studies demonstrating that multimodal fusion strategies with histopathological images and genomic profiles improved clinical prediction and patient stratification over digital pathology and molecular methods alone. The authors also discussed the development of pan-cancer and multimodal models across multiple cancer types, showing improved predictive performance compared to unimodal approaches.

Have you heard of minimal residual disease testing? 🤔 💠 Minimal residual disease (MRD) testing is a tool that is used for finding any remaining disease following cancer treatment, monitoring response to treatment, and to guide potential adjustments to the care plan. 💠 MRD testing first achieved FDA approval for use in blood cancers, such as leukemia and lymphoma. 💠 MRD has grown in use in more recent years for solid tumors, such as breast cancer. ✔ One of the advances in cancer care is the ability to detect Circulating Tumor DNA via minimal residual disease testing. 💡 Circulating tumor DNA (ctDNA) is a DNA fragment actively secreted by tumor cells or released into the circulatory system during the process of apoptosis or necrosis of tumor cells. 💡 Frankly, this all gets a tad complex, but the takeaway is that ctDNA testing is a minimally invasive technique that can be used to characterize individual cancer biology and monitor disease. 💡 ctDNA testing is performed via a liquid biopsy - A blood sample. Obtaining blood sample is safe, inexpensive, and easy to repeat. 💡 ctDNA can detect the recurrence of cancer before it is found in imaging, such as PET/CT. 💡 The European Society Of Medical Oncology (ESMO) recommendations endorse ctDNA testing in routine clinical practice for tumor genotyping to direct molecularly targeted therapies in patients with metastatic cancer. 📝 ctDNA technologies are still being investigated, but preliminary research has demonstrated that it can be a sensitive and specific approach to breast cancer surveillance of disease recurrence. 💠 At our breast cancer clinic, we frequently utilize ctDNA tests manufactured by Guardant, Tempus, and Signatera. I think it is an exciting development in oncology treatment and I look forward to further research on its efficacy! References: 📙 National Cancer Institute 📘 Journal of Clinical Oncology 📗 Nature Partner Journals 📔 European Society Of Medical Oncology (ESMO) _____________________________________________________________________ Charlotte Goor – 🩺Registered Nurse. ⚖Legal Nurse Consultant. 📧 Charlotte@ExpertCareLNC.com 💻http://ExpertCareLNC.com

Today, on World Cancer Day, we recognize the profound impact cancer has on individuals and families worldwide. My father had stage IIIB adenocarcinoma of the lung, with his left upper lobe removed, and my uncle succumbed to small cell lung cancer. Both were non-smokers. These stories underscore the urgency of advancing our detection methods. It's a personal mission for many, driven by the hope that through technology, particularly the fusion of Knowledge AI and Big Data AI, we can unveil these silent killers early enough to make a difference. Here's a proposed 10-step protocol for deploying an algorithm capable of early detection of solitary lung nodule cancer, leveraging blood biomarkers, radiology, and other modalities: Data Collection and Integration: Gather extensive datasets covering various patient demographics and stages of lung cancers. Big Data Infrastructure: Develop efficient data handling for structured and unstructured data. Knowledge AI Models: Utilize medical knowledge to enhance AI models. Machine Learning and Deep Learning: Apply AI techniques for identifying early-stage cancer patterns. Radiology Image Analysis: Train AI for advanced image recognition of lung scans. Blood Biomarker Detection: Develop algorithms for non-invasive blood test analysis. Predictive Modeling: Personalize risk assessments using predictive models. Clinical Validation: Ensure model accuracy through extensive clinical trials. Integration into Clinical Workflows: Collaborate with healthcare providers to incorporate AI into existing processes. Continuous Learning and Improvement: Establish a system for regular AI model updates based on new data and discoveries. By following these steps, we can harness AI's power to transform early lung cancer detection, potentially saving countless lives. The fusion of Knowledge AI and Big Data AI offers hope, turning silent stories into beacons of progress. Through early detection, we aspire to beat cancer.

Can You Hear Me Now? Might ctDNA Hear Cancer Before It Shouts? As an oncologist, I've witnessed firsthand the profound impact of medical advancements. The prospect of "hearing" cancer's earliest molecular whispers through circulating tumor DNA (ctDNA) before it "shouts" through symptoms is undeniably one of the most exciting frontiers in our field. This technology promises a future where we might intercept cancer far earlier and manage it more precisely. 🔬 The Dawn of Molecular Listening: Tools like Multi-Cancer Early Detection (MCED) tests hope to identify many cancers from a single blood draw, potentially transforming screening paradigms. Similarly, Minimal Residual Disease (MRD) testing is already helping us personalize post-treatment care for some cancers, offering a clearer view of what might remain after initial therapy. Much like the precise molecular monitoring achieved in Chronic Myeloid Leukemia (CML), the aspiration is to bring this clarity to a broader range of cancers. 🩺 Balancing Pioneering Hope with Prudent Care: The potential is immense, and for our patients, especially those at high risk like BRCA carriers or individuals anxiously monitoring for recurrence post-surgery, these developments spark understandable hope. They see a "window of opportunity" – a chance to act decisively at the faintest signal. As physicians, we share that desire for progress. Yet, our foremost commitment is to "first, do no harm." We must approach these powerful new tools with optimism and a sober, meticulous commitment to evidence. We must rigorously evaluate peer-reviewed data and validated results, ensuring that any new diagnostic or intervention benefits our patients without undue risk or false promise. This inherent tension between population-based evidence and individual hope, between 'do no harm' and 'miss no chance,' lies at the heart of integrating these disruptive, powerful technologies into compassionate cancer care. It's a conversation we navigate daily with our patients, weighing the established benefits and risks against the potential of emerging science. ✨ The Path Forward: Responsible Innovation: The journey to fully integrate ctDNA technologies requires continued rigorous research, transparent data reporting (successes and limitations), and thoughtful ethical consideration. Our collective goal must ensure these molecular insights translate into genuinely improved outcomes – more lives saved, better quality of life, and true peace of mind. This transformation is happening and calls for careful navigation from all of us in the healthcare community. Please take a look at my latest piece for a deep dive into these questions. #CancerCare #ctDNA #LiquidBiopsy #EarlyDetection #PatientAdvocacy #Oncology #PrecisionMedicine #EvidenceBasedMedicine #HealthcareInnovation

Digital Health in Health Assessment & Medical Diagnostics Global Startups Landscape 2.5Q 2024 is evolving rapidly, driven by AI-powered innovations across various medical fields. AI-Driven Diagnostics: AI is at the forefront of medical diagnostics. Startups like Aiberry (mental health) and Aidoc (cardiovascular health) are using AI to analyze data in real-time, improving early diagnosis and decision-making. These technologies offer non-invasive, faster, and more accurate assessments than traditional methods. Medical Imaging and Radiology: AI-powered imaging is a key area, with startups like Aidence (lung cancer screening) and Paige (digital pathology) leading the way in enhancing radiological diagnostics. These companies are pushing the boundaries of precision medicine, improving early detection and workflow efficiencies for radiologists and pathologists alike. Portable and Wearable Devices: Portable and wearable diagnostic tools are gaining prominence, exemplified by Butterfly Network, Inc. (handheld ultrasound) and Hyperfine, Inc. (portable MRI). These startups are making high-quality medical imaging more accessible, especially in underserved regions. Predictive and Personalized Medicine: Companies like Cardiosense (cardiovascular health) and Freenome (cancer detection) are leveraging multi-sensor devices and AI to predict disease onset, providing personalized treatment recommendations. This shift toward predictive healthcare is reshaping patient care, enabling more proactive intervention strategies. Voice and Speech Biomarkers: In mental health, companies like Sonde Health, Inc. and Kintsugi are innovating by using voice technology to detect signs of depression and anxiety, proving the versatility of AI in mental health diagnostics and offering real-time mental health assessments. Women’s Health: LEVY Health (endocrine disorders and fertility), Sonio (prenatal ultrasound), and Nevia bio (early disease detection) are advancing women’s health diagnostics, focusing on reproductive and prenatal health through AI-powered decision support platforms. Cross-Specialty Diagnostics: Startups such as Viz.ai and PathAI provide cross-specialty diagnostic tools, focusing on synchronizing care in fields like neurology and pathology. Viz.ai facilitates faster stroke care with its AI-driven platform, whereas PathAI uses AI to enhance diagnostic accuracy in pathology, especially in cancer diagnostics. Global startups in this space are attracting significant investments, with companies like Aidoc raising substantial funds to expand their platforms to more conditions and regions. Achieving CE marking and FDA clearances, as seen with companies like Ultromics, is essential for global expansion and validation. #DigitalHealth #Healthcare #Assessment #Medical #Diagnostics #AIinHealthcare 

Brain Cancer Has a New Look: How Pathology Is Going Multimodal On a recent Saturday morning, I was called in to review a brain tumor intraoperatively. We didn’t do a frozen section. We haven’t in years. Instead, we reviewed a stimulated Raman spectroscopy image—a laser-based, label-free technique that produces near real-time images from fresh tissue, without freezing or staining. It’s non-destructive and fast enough to guide surgical decision-making. At our institution, we were the first in the U.S. to formally replace frozen sections for brain tumors with this technology. Not as a supplement—as the standard of care. Everyone benefits: • The surgeon gets an answer in minutes. • The pathology team isn’t tied up at the cryostat. • The patient spends less time under anesthesia and gets more diagnostic clarity. This is a glimpse of where pathology is going. Just as radiology evolved from plain films to CT, MRI, and PET, pathology is becoming multimodal. Optical microscopy, ex vivo and in vivo imaging, multispectral visualization, genomics, and AI—all of it expanding how we see, interpret, and diagnose disease. The microscope isn’t going away. But it’s no longer the only tool in the toolbox. Invenio Imaging Robert Louis, MD, FAANS, FCNS #pathology #digitalpathology #neurosurgery #imaging #innovation

🚀 Diagnostics & Digital Pathology Are Colliding. And Tempus AI is leading the charge. Having spent most of my career in Diagnostics, particularly in Genetics & Genomics, before diving into Digital Pathology, I’ve seen these two worlds operate in silos for years. Genomics interprets DNA. Pathology analyzes tissue. But the future of diagnostics isn’t either/or, it’s both. And Tempus is making that a reality. As Chris Scotto DiVetta put it: "Tempus is committed to broadly offering AI-enabled diagnostics to improve patient outcomes." Tempus is doubling down on AI-powered, multi-modal diagnostics with two recent partnerships: 1️⃣ ArteraAI - AI-powered prostate cancer diagnostics - The NCCN-backed & Medicare-covered ArteraAI Prostate Test uses AI to analyze biopsy images and predict patient outcomes & therapy response. - For oncologists, this means faster, smarter treatment decisions. 2️⃣ Imagene - AI-driven lung cancer biomarker prediction - Imagene’s AI detects key NSCLC biomarkers directly from biopsy in minutes. - Integrated with Tempus’s genomic & clinical data, this could revolutionize lung cancer diagnostics. 📊 The Real Play? Data. In my opinion, Tempus isn’t just partnering with AI companies, it’s building data pipelines that strengthen its dominance in AI-driven diagnostics. 💡 Tempus’s Data & Services grew 45%, far outpacing its testing business. 💡 The high-margin data-driven revenue boosted profits by nearly 50%. Similarly, I believe the Ambry Genetics acquisition (while expanding the testing menu, especially in germline) was another data play, fueling Tempus’s precision medicine intelligence engine. 🔹 Why these AI partnerships matter: - Every Imagene biomarker prediction = More pathology & genomic data. - Every ArteraAI prostate test = More clinical outcome data. 𝐓𝐞𝐦𝐩𝐮𝐬 𝐢𝐬 𝐭𝐡𝐞 𝐀𝐈-𝐩𝐨𝐰𝐞𝐫𝐞𝐝 𝐝𝐚𝐭𝐚 𝐢𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞 𝐜𝐨𝐦𝐩𝐚𝐧𝐲 𝐨𝐟 𝐩𝐫𝐞𝐜𝐢𝐬𝐢𝐨𝐧 𝐦𝐞𝐝𝐢𝐜𝐢𝐧𝐞. Tempus is solidifying its position as a leader in precision diagnostics, integrating genomic sequencing, AI pathology, and clinical informatics. In a rapidly evolving field, scale & integration are crucial and Tempus is making moves to stay in the lead. More M&A on the Horizon? Is Tempus hinting at its next acquisitions? 💬 Which #DigitalPathology or #Diagnostics company do you think will be next to partner with, or be acquired by a major player like Tempus?