Broad Institute of MIT and Harvard’s Post

If you want to learn more about a novel deep learning-based multimodal clustering model that integrates complex multi-fluid, multi-omic data to assist in KOA patient endotyping and test outcome response to TKA surgery, then this is a must-read. The study includes 2,727 molecular features from four domains: 151 plasma metabolites, 421 plasma miRNAs, 930 synovial fluid miRNAs, and 1,225 urine miRNAs. We were able to unravel the heterogeneity of a sample of late-stage surgical KOA patients and evaluate post-TKA response classification. Congratulations to the entire team for their endless support and dedication! Your hard work and collaboration have been instrumental in achieving this milestone. #Deeplearning #miRNASequencing #Metabolomics #Illumina

This deep learning model allowed estimation of FVC and FEV1 from chest x-rays, showing high agreement with spirometry. The model offers an alternative to spirometry for assessing pulmonary function, which is especially useful for patients who are unable to undergo spirometry, and might enhance the customisation of CT imaging protocols based on insights gained from chest x-rays, improving the diagnosis and management of lung diseases.

🎙️ Publication Alert Our newly published study examined the effectiveness of various predictive models for time-to-event outcomes for predicting cardiovascular disease (#CVD) risk in healthy older adults. 🖥️ 💻We focused on machine learning (#ML) and deep learning (#DL) survival models, comparing them to the standard survival model. Our findings indicate that #DL models are more effective than #ML and the standard survival models in predicting the 10-year risk of CVD. 🪜 Our study adds to the growing body of evidence supporting the use of advanced #AI methodologies in healthcare, particularly in enhancing the accuracy of cardiovascular risk predictions. Thanks to the authors - Dr Htet Lin Htun, Dr Mor Vered, Dr Alice Owen, Dr Joanne Ryan, Professor Andrew Tonkin, and Dr Rosanne Freak-Poli To access the paper 👉https://lnkd.in/g8dFXzN5

Just published by my group our work "Addressing hidden risks: Systematic review of artificial intelligence biases across racial and ethnic groups in cardiovascular diseases." As AI becomes integral to cardiovascular medicine, a crucial question arises: How robust are these models when applied outside the (often very small) populations in which they are developed and validated? How do algorithms, trained on specific populations, perform when used in entirely different ones? How generalizable are their results? Addressing racial and ethnic disparities is essential to ensure equitable healthcare for all. In this study, we explored these issues, and the findings are striking: 82% of the studies revealed significant disparities in AI model performance across racial and ethnic groups. Takeaway: Pay close attention to the AI tools you use and how you use them! https://lnkd.in/e2HKgMNn

"A machine-learning study at Weill Cornell Medicine was able to classify Parkinson’s disease into three subgroups, a development with the potential to effectively target patients with treatments specific to their disease’s progression." #ai #artificialintelligence #machinelearning #research #parkinsonsdisease #deeplearning

Can AI models improve the accuracy of CVD risk prediction? TL;DR - yes! 😄 In this paper, we showed that deep learning models such as DeepSurv and NMTLR had better precision compared to standard Cox model.

Researchers at USF Health and Weill Cornell Medicine, as part of an expansive, multi-institutional project investigating voice as a biomarker for disease, have reached a significant milestone by publishing the first version of their clinically validated voice dataset to an online artificial intelligence platform where it will be an invaluable resource for researchers across the globe. From our web story: "Artificial intelligence is revolutionizing our ability to detect and understand disease, and this groundbreaking voice dataset is a monumental step forward in that journey," said Dr. Olivier Elemento, Director of the Englander Institute for Precision Medicine, who is also a professor of physiology and biophysics at Weill Cornell Medicine. "These clinically validated data, combined with cutting-edge AI techniques, pave the way for new diagnostic possibilities and groundbreaking innovations that will transform patient care globally." #AI #PrecisionMedicine https://lnkd.in/esa69PqV

🌟 Exciting Research Announcement! 🌟 In the era of rapid digital transformation and amidst the unprecedented challenges posed by the COVID-19 pandemic, the imperative for accurate and efficient disease diagnosis has never been more pressing. I am thrilled to share that our latest research paper on the automated detection of two critical respiratory illnesses, COVID-19 and pneumonia, has been published! 🚀 In this study, we employ a sophisticated approach integrating three distinct models: a custom model developed in-house, Xception, and DenseNet121. Leveraging the power of Convolutional Neural Networks (CNNs), renowned for their exceptional capabilities in pattern recognition within medical images, our research provides a thorough evaluation of each model's performance. Through meticulous experimentation and rigorous comparative analysis, we elucidate not only the strengths and limitations of these models but also their potential for practical deployment in clinical settings. By juxtaposing the outcomes derived from our custom model with those from the established architectures of Xception and DenseNet, we offer a nuanced understanding of their respective efficacies in disease detection. This study aspires to contribute substantively to the ongoing discourse in medical image analysis, aiming to enhance disease detection accuracy and ultimately improve patient care outcomes amidst the challenges posed by respiratory illnesses such as COVID-19 and pneumonia. I am incredibly proud of this work and look forward to its impact on the field. A huge thank you to my team and collaborators for their dedication and support. 🙏 Sarvesh Chaudhari Ankush Jain SAURABH HUNDARE 📝 Research Paper Link: https://lnkd.in/diY5rzES 🎥 Project Presentation: https://lnkd.in/dBJgg2T3 #Research #MedicalImaging #COVID19 #Pneumonia #DeepLearning #CNN #HealthcareInnovation #ArtificialIntelligence #MedicalResearch

🌟 AI-Powered Tongue Color Analysis: A Breakthrough in Real-Time Disease Diagnosis 🌈🔬 🚀 Innovative Disease Detection: Researchers from Middle Technical University (MTU) and the University of South Australia (UniSA) have achieved a significant advancement in medical diagnostics with an AI model that analyzes tongue color to predict diseases with 98% accuracy. This breakthrough technology promises to revolutionize how we diagnose conditions such as diabetes, COVID-19, anemia, and more. 🌟🩺 🤖 AI in Action: -- Precise Predictions: The AI model uses advanced machine learning algorithms to analyze the color of the tongue and identify a range of health conditions. The technology draws on traditional Chinese medicine practices, where tongue examination has long been used to detect various diseases. 📈🧬 -- Disease Detection: The AI system can diagnose conditions including diabetes (yellow tongue), stroke (unusually shaped red tongue), anemia (white tongue), and severe COVID-19 (deep red tongue), among others. It also identifies signs of vascular and gastrointestinal issues and asthma based on tongue color variations. 🌡️🔍 📊 Technical Details: -- Comprehensive Training: The AI was trained using 5,260 images of tongues across different color classes (red, yellow, green, blue, gray, white, and pink) using six machine-learning algorithms: naïve Bayes, support vector machine, k-nearest neighbors, decision trees, random forest, and Extreme Gradient Boost. This diverse training set allows the model to work effectively under various lighting conditions. 🖼️💻 🔬 Real-Time Diagnosis: -- Immediate Results: By analyzing tongue colors in real-time, the AI model offers rapid disease detection and diagnosis, significantly speeding up the diagnostic process. The technology has already demonstrated success in matching tongue colors with diseases in patient images from teaching hospitals. ⚡🏥 🌍 Implications for Healthcare: -- Enhanced Diagnostic Speed: This technology could transform patient care by providing quick and accurate disease predictions, potentially leading to earlier interventions and improved health outcomes. 🌟🩺 -- Accessible Screening: The use of AI for tongue color analysis makes it possible to screen for a variety of conditions efficiently and affordably, offering a valuable tool for healthcare providers worldwide. 🌐💡 Conclusion: The AI-driven tongue color analysis represents a major leap forward in disease diagnosis, leveraging advanced machine learning to offer precise, real-time predictions. This innovation not only enhances diagnostic accuracy but also promises to improve healthcare accessibility and patient care. 🌟🔬🩺 #AIInHealthcare #DiseaseDetection #TongueColorAnalysis #MedicalInnovation #RealTimeDiagnosis #HealthcareTechnology #MachineLearning #FutureOfMedicine #HealthTech https://lnkd.in/eGrZR7Gg

Can #lung function be accurately determined from a simple chest X-ray using an #artificial #intelligence (#AI) algorithm? Japanese researchers explored this question in a recent study, analyzing over 140,000 chest X-rays paired with spirometry results from more than 80,000 patients across five institutions. They found a strong correlation between the AI algorithm's estimates of FEV1 and FVC from chest X-rays and the actual spirometry measurements. The correlation coefficients were promising. The study population had a low to moderate incidence of pulmonary diseases, but patients with conditions like #COPD, #asthma, #ILD, #lungcancer, and #tuberculosis comprised 1% to 45% of the participants, depending on the institution. This study is intriguing and suggests that AI could potentially estimate lung function from imaging, especially for patients unable to undergo spirometry. It may also serve as a tool for personalized pulmonary investigations, such as selecting cases for lung function testing or other imaging tests. However, further research is needed to validate this algorithm in other populations, including those outside Japan and those with a higher incidence of #pulmonary disease