🚨 New in our Signal Processing in Medicine series (#10): A massive new EEG dataset enables real-time decoding of imagined speech from non-invasive brain signals, the Chinese Imagined Speech Corpus (Chisco) with over 20,000 sentences and 900 minutes per participant. Published in Scientific Data (Nature, 21 November 2024). Advanced signal processing machine learning now brings us closer to practical thought-to-text BCI without implants. This is non-invasive signal processing unlocking natural brain-computer communication. 👉Poll: Would you use a non-invasive EEG headset AI to turn your thoughts into text in real time? A) Yes — revolutionary for communication B) Only for medical use (e.g. locked-in patients) C) Not yet — privacy & accuracy concerns D) Other → comment! Vote reply! #BiomedicalSignalProcessing #MedicalAI #BCI #EEG #NatureScientificData #NeuroTech

🚨 📌 Pinned: Recent Signal Processing in Medicine Series Index Catch every breakthrough in biomedical signal processing AI (cardiology, neurology, neurotech & wearables). Here are the latest installments (#5–#9): #9 Edge-AI Closed-Loop Implants for Epilepsy Real-time seizure detection neurostimulation on neuromorphic chips (ultra-low power, sub-ms response). Published: PNAS Nexus 🔗 Full thread paper poll: x.com/AyyoubAkbari/status/20… #8 Wearable ECG PPG for Real-Time Anxiety Detection Smartwatch signals achieve clinical-grade accuracy (up to 92%) rivaling structured interviews. Published: npj Digital Medicine 🔗 Full thread paper poll: x.com/AyyoubAkbari/status/20… #7 Multimodal LLMs That “Read” 12-Lead ECG Images Open-source PULSE 7B outperforms GPT-4o, Claude 3.5 Sonnet & Gemini by 21–33%. Published: npj Digital Medicine 🔗 Full thread paper poll: x.com/AyyoubAkbari/status/20… #6 Device-Agnostic Cardiac Foundation Model (CSFM) One multimodal model works across ANY ECG/PPG device (1.7M patients pretrained). Published: Nature Machine Intelligence 🔗 Full thread paper poll: x.com/AyyoubAkbari/status/20… #5 EEG-Based BCI for Individual Finger-Level Robotic Hand Control Fine-grained motor imagery decoding (80% accuracy on 2–3 finger tasks). Published: Nature Communications 🔗 Full thread paper poll: x.com/AyyoubAkbari/status/20… Follow @AyyoubAkbari for weekly deep-dive threads, open-access papers, practical code, polls & the next installment in the series. Med-tech, #cardiology , #neurology , #neuroengineering & IEEE folks, this is for you. Bookmark • Retweet • Turn on notifications #BiomedicalSignalProcessing #MedicalAI #NeuroTech #ECG #EEG #Wearables #IEEE

🚨 New in our Signal Processing in Medicine series (#9): Edge-AI on implantable devices can now detect seizures in real time and deliver closed-loop neurostimulation to abort them, published in PNAS Nexus (30 Oct 2024). Neuromorphic chips on-device signal processing achieve ultra-low power (pJ per spike) and sub-millisecond response, turning intracranial EEG into autonomous, adaptive therapy. No more constant doctor intervention. Precision epilepsy treatment via embedded signal processing is here. Poll: Would you support fully autonomous closed-loop implantable AI for drug-resistant epilepsy? A) Yes, life-changing for patients B) Only with strict human oversight C) Not yet, need more long-term safety data D) Other → comment! Vote reply! #BiomedicalSignalProcessing #MedicalAI #Epilepsy #NeuroTech #PNASNexus

🚨 New in our Signal Processing in Medicine series (#7): Multimodal LLMs can now “read” 12-lead ECG images like a cardiologist — the open-source PULSE model (7B parameters) outperforms Claude 3.5 Sonnet, GPT-4o, and Gemini by 21–33% on real clinical benchmarks. Published in npj Digital Medicine (16 March 2026). Trained on over 1 million synthetic real ECG images (ECGInstruct dataset) with realistic artifacts, perfect for printed reports or resource-limited settings. Vision-language AI meets traditional signal processing. Poll: Have you tried (or would you try) an LLM directly on 12-lead ECG images for a second opinion? A) Yes — huge time-saver B) Only if fully validated by cardiologists C) Not yet — prefer raw signal analysis D) Other → comment! Vote reply! #BiomedicalSignalProcessing #MedicalAI #ECG #LLM #npjDigitalMedicine

🚨 New in our Signal Processing in Medicine series (#6): A single multimodal foundation model (CSFM) pretrained on 1.7 million individuals now assesses cardiac health across ANY device, ECG (12-lead or single-lead), PPG, or mixed, published in Nature Machine Intelligence (24 Feb 2026). Transformer-based generative masked pretraining creates unified representations that outperform single-task models on diagnosis, vital signs, risk prediction, and clinical reporting. No more retraining per sensor. This is signal processing at massive scale. Poll: Would you use a single AI foundation model for all your cardiac signals (ECG PPG from any wearable)? A) Yes — game changer for clinics B) Only after full FDA approval C) Not yet — prefer device-specific tools D) Other → comment! Vote reply! #BiomedicalSignalProcessing #MedicalAI #ECG #PPG #NatureMachineIntelligence

🚨 New in our Signal Processing in Medicine series (#5): EEG-based brain-computer interfaces just achieved individual finger-level robotic hand control, published in Nature Communications (30 June 2025). Using motor execution & motor imagery, the system decodes fine-grained brain signals in real time with deep neural networks (EEGNet) advanced filtering (FBCSP, alpha/beta band ERD). Accuracies: 80.56% (2-finger MI) and 60.61% (3-finger MI) in experienced users, a major leap toward naturalistic prosthetic control. #BiomedicalSignalProcessing #EEG #BCI #MedicalAI #NatureCommunications #NeuroTech

🚨 As promised in the Claude NeuroKit2 thread. here’s the exact prompt ready-to-run code: Prompt you paste into Claude: “Using NeuroKit2, write clean Python code to: load a sample 24-hour ECG .edf or .csv file, remove motion artifacts, detect R-peaks, compute HRV metrics (SDNN, RMSSD, HF/LF), and identify AF/PVCs. Include clean matplotlib plots and a clinical-style summary.” Claude instantly returns production-ready code that works out of the box. #BiomedicalSignalProcessing #MedicalAI #ECG #NeuroKit2 #ClaudeAI #Medical #AI #SignalProcessing

🚨 Game-changing biomedical signal processing breakthrough that could prevent thousands of strokes: AI now analyses simple single-lead ECGs (handheld or smartwatch-style) to predict future atrial fibrillation (AF) risk with accuracy matching traditional clinical scores — published in npj Digital Medicine (26 Nov 2025). No more age-only screening. Advanced filtering AI turns noisy real-world signals into reliable early warnings. This is preventive cardiology powered by signal processing. Would you trust a smartwatch or handheld ECG AI to screen you for hidden AF risk? A) Yes — early detection saves lives B) Only with doctor oversight C) Not yet — need more proof D) Something else → comment! Vote reply! #BiomedicalSignalProcessing #MedicalAI #AFib #ECG #AIinHealthcare

Quick follow-up to yesterday’s #Claude NeuroKit2 thread 👇 The exact prompt ready-to-run code is coming in the next post (as promised). If you work in cardiology, med-tech or biosignal processing, this series is for you, follow for more! What’s one biosignal task you’d love #AI to automate? #BiomedicalSignalProcessing #MedicalAI @IEEEorg @NatureComms @AnthropicAI @claudeai

🚨 New in our Signal Processing in Medicine series: Claude AI is now a real-time collaborator for biomedical signal processing. Real clinical scenario (2026): A cardiologist receives noisy 24-hour ambulatory ECG data from a wearable device — motion artifacts everywhere, thousands of beats. They open Claude (with its NeuroKit2 integration), paste a short prompt sample data, and instantly get: → Clean Python code to load, filter, segment, and analyse the full recording → HRV metrics, arrhythmia detection (AF, PVCs, VT) → Publication-ready plots clinical summary #BiomedicalSignalProcessing #MedicalAI #ECG #ClaudeAI #AIinHealthcare

🚨 Cutting-edge biomedical signal processing breakthrough in Nature Communications [02 April 2025]: Machine learning supercharges point-of-care vertical flow assays (xVFA / hs-VFA). Au-ion amplification fully connected neural networks (FCNN) turn portable paper tests into lab-grade quantitative tools. Detects high-sensitivity cardiac troponin I (hs-cTnI) at 0.2 pg/mL (10× below clinical cutoff) in minutes — with Pearson’s r > 0.96 correlation to FDA-approved analyzers and <6.2% CV repeatability. No central lab needed. Real impact for emergency cardiac diagnostics. #BiomedicalSignalProcessing #MedicalAI #POCT #NatureCommunications

🔬 Bioengineering's Hot Special Issues Collection on Biomedical Signal Processing is live; AI diagnostics, biosignals, imaging, gait, fetal monitoring & more. All open access! 🔗 brnw.ch/21x2dc5 #BiomedicalSignalProcessing #AIinHealthcare #BiomedicalImaging

🔬 Bioengineering's Hot Special Issues Collection on Biomedical Signal Processing is live; AI diagnostics, biosignals, imaging, gait, fetal monitoring & more. All open access! 🔗 brnw.ch/21x2dc5 #BiomedicalSignalProcessing #AIinHealthcare #BiomedicalImaging

Peak Attention U-Net: Enhancing ECG delineation with attention #UNet #ECGAnalysis #AttentionMechanism #BiomedicalSignalProcessing #DeepLearning International Robotics and Automation Awards Visit Us :roboticsandautomation.org Nomination:roboticsandautomation.org/aw…

The Department of Biotechnology & Medical Engineering, NIT Rourkela, is organizing a five-day online short-term program on “Recent Advances in Biomedical Signal and Image Processing” from 23rd to 27th February 2026. Registration is open to PG students, Ph.D. scholars, Post-Doctoral Fellows, and faculty members of engineering institutions. The last date for registration is 20th February 2026. 🔗Registration link: forms.gle/mCzFFZ55wgzTw5dh9 The program focuses on recent developments in biomedical signal and image processing, with hands-on exposure to AI-based diagnostic tools, ECG and EEG analysis, medical image enhancement, deep learning applications in healthcare, and emerging trends in biomedical research and clinical practice. The program is coordinated by Dr. J. Sivaraman and Dr. Bala Chakravarthy N. Brochure: nitrkl.ac.in/docs/CEP/BM/290… #NITRourkela #BiomedicalEngineering #BiomedicalSignalProcessing #MedicalImageProcessing #ShortTermProgram #ResearchAndInnovation #HealthcareTechnology

⚡ Epilepsy Signal Processing – May 30 How can signal processing help epilepsy surgery? Join Dr. Vasileios Kokkinos' invited talk at SIPPRE/EUNICE 🗓 May 30, 16:00 GR 🔗 Register: events.teams.microsoft.com/e… #Epilepsy #Neuroengineering #BiomedicalSignalProcessing #SIPPRE

IoT-Based Heartbeat Rate-Monitoring Device Powered by Harvested Kinetic Energy mdpi.com/1424-8220/24/13/424… #biomedicalsignalprocessing

Wireless Mouth Motion Recognition System Based on EEG-EMG Sensors for Severe Speech Impairments mdpi.com/1424-8220/24/13/412… #biomedicalsignalprocessing #wearable #biomedicalsensors

Welcome to read and share the Highly Accessed Article in 2023. 📢 Title: Transfer Learning Approach for Human Activity Recognition Based on Continuous Wavelet Transform 📢 Paper link: mdpi.com/1999-4893/16/2/77 #biomedicalsignalprocessing #humanactivityrecognition #CNN

QRS Detector Performance Evaluation Aware of Temporal Accuracy and Presence of Noise mdpi.com/1424-8220/24/5/1698 #electrocardiogram #ECG #biomedicalsignalprocessing