Chunli Shao

Arrhythmia poses a severe health risk, necessitating long-term monitoring to prevent complications like stroke. However, existing radar monitoring solutions fail to detect concurrent arrhythmias and lack clinical interpretability. In this paper, we bridge the gap by proposing READ (Radar Evidence-based Arrhythmia Diagnosis). Unlike previous “black-box” approaches, READ integrates mmWave radar with Large Language Models (LLMs) to deliver transparent, evidence-based diagnoses. However, training such a reasoning-capable model is hindered by the scarcity of mmWave datasets with detailed diagnostic annotations. To tackle this challenge, we first pretrain on large-scale ECG data to learn robust physiological patterns and then transfer this knowledge to radar domain using our proposed Cross-Physics Knowledge Distillation (CPKD) framework. Furthermore, we design a Morphological Grouping Tokenizer and a Morphology-Anchored Clinical Reasoning module, which parse raw signals into semantic tokens to guide step-by-step diagnostic inference. Empirical evaluation on a real-world dataset of 205 patients, collected in collaboration with medical institutions, demonstrates READ 's superior performance, achieving a 90.12% exact match ratio and a 0.9625 F ₁ score. Our framework significantly outperforms state-of-the-art baselines while providing interpretable clinical reports.

Heart sound, as an important physiological signal representing the state of the cardiovascular system, is crucial for the diagnosis of cardiovascular disease (CVD). Traditional heart sound detection methods rely on doctors’ experience and specialized auscultation equipment, but these methods have limitations in terms of efficiency and environmental adaptability. To address these issues, this article proposes a mobile phone-based heart sound detection method. It uses the built-in microphone to record heart sounds and adopts a two-stage filtering method to improve the quality of heart sounds. Next, we design a multifeature fusion network named MBCF-Net to extract multidomain features, and effectively cross-domain fuse them while enhancing important channel information. We evaluate the proposed model on a self-collected smartphone heart sound dataset and two public heart sound datasets. The experimental results show that MBCF-Net achieves classification accuracies of 99.63%, 98.18%, and 99.88% on the smartphone dataset, the PhysioNet/CinC Challenge 2016 dataset, and Yaseen’s five-class dataset, respectively. Compared with previous works, MBCF-Net achieves higher diagnostic accuracy, demonstrating its strong performance and adaptability across different heart sound classification scenarios.

The utilization of wearable devices with integrated photoplethysmography (PPG) sensors for continuous and comfortable Arterial Blood Pressure (ABP) monitoring has gained popularity. However, our analysis on a large-scale self-collected dataset reveals a significant challenge: approximately 56.7% of individuals experience the absence of the dicrotic notch (a vital morphology feature) in wrist-PPG data, while 23.8% face the same issue in finger-PPG data, which poses a big obstacle to generalized ABP monitoring. To address this challenge, we propose a morphology-independent ABP estimation algorithm based on the Frank-Starling law. We also devise techniques to address calibration and PPG data drift issues commonly faced in continuous ABP monitoring. Further, a system called RingBP is developed towards continuous, comfortable, and generalized ABP monitoring using a smart ring. We design and implement a smart ring prototype and conduct extensive experiments involving 85 participants, demonstrating its superiority over other solutions.

Background: Despite the World Health Organization's prioritization of familial hypercholesterolemia (FH), its global diagnostic rate remains critically low, leading to inadequate treatment and control, thereby increasing the risk of atherosclerotic cardiovascular disease. This study aimed to investigate the comorbidity burden of FH in China and analyze the differences between familial and general hypercholesterolemia (HC) populations.Methods: Using a national medical insurance database from 2013 to 2017 including 13,976 patients with FH and 13,976 matched control patients with HC, we utilized case‐control methods to compare the composition ratio, comorbidity rates, medical expenses, and healthcare burden of patients with FH to those of control patients.Results: The FH population had a higher comorbidity rate of more than one cardiometabolic disease (83.7% [11,697/13,976]) compared to the HC group (70.3% [9279/13,976]; χ² = 250.45, p < 0.0001). The rates of coronary heart disease, hypertension, stroke, and diabetes were higher in patients with FH (39.2% [5475/13,976], 71.0% [9925/13,976], 14.2% [1982/13,976], and 31.2% [4363/13,976], respectively) compared to those in the HC group (30.4% [4255/13,976], 61.4% [8587/13,976], 11.5% [1601/13,976], and 28.1% [3923/13,976], respectively; all p < 0.0001). In the 40−49 age group, patients with FH had a significantly higher average number of comorbidities compared to control patients with HC (1.2 vs. 0.9; t = 15.67, p < 0.0001). Notably, the comorbidity count in patients with FH aged 40−49 years even exceeded that in patients with HC aged 50−59 years. Furthermore, the annual per capita medical cost for patients with FH was significantly higher at 5045.5 Chinese yuan (CNY) compared to 4184.7 CNY for patients with HC (t = 12.54, p < 0.0001).Conclusion: With a large number of patients with dyslipidemia, the type and number of comorbidities significantly impact the healthcare burden. FH presents with earlier onset, more comorbidities, and heavier cardiovascular‐related medical burdens than HC. Early identification, intervention, and comprehensive management of comorbidities in the FH population are crucial for neuroprotection and prevention of atherosclerotic cardiovascular disease.

Large language models (LLMs) are rapidly advancing medical artificial intelligence, offering revolutionary changes in health care. These models excel in natural language processing (NLP), enhancing clinical support, diagnosis, treatment, and medical research. Breakthroughs, like GPT-4 and BERT (Bidirectional Encoder Representations from Transformer), demonstrate LLMs’ evolution through improved computing power and data. However, their high hardware requirements are being addressed through technological advancements. LLMs are unique in processing multimodal data, thereby improving emergency, elder care, and digital medical procedures. Challenges include ensuring their empirical reliability, addressing ethical and societal implications, especially data privacy, and mitigating biases while maintaining privacy and accountability. The paper emphasizes the need for human-centric, bias-free LLMs for personalized medicine and advocates for equitable development and access. LLMs hold promise for transformative impacts in health care.

Background: It remains uncertain whether a novel composite indicator, the high sensitivity C‐reactive protein (CRP) (hsCRP)‐to‐body mass index (BMI) ratio (CBR), is associated with unfavourable events in subjects who have undergone drug‐eluting stent (DES) implantation. Therefore, we conducted a prospective cohort study to evaluate the associations between the CBR and adverse outcome occurrences. Methods: From January 2013 to December 2013, a total of 9810 subjects who underwent DES implantation were enrolled in the study. The participants were divided into three groups on the basis of the CBR. We defined the primary endpoint as the occurrence of major adverse cardiovascular and cerebrovascular events (MACCEs), and the secondary endpoints included the occurrence of all‐cause death, myocardial infarction(MI), stroke or target vessel revascularization (TVR). Kaplan‒Meier (K‒M) analysis and Cox regression analysis were performed to assess the discrepancy in the risk of adverse outcomes between the different CBR groups. Results: Over an average follow‐up period of 26.7 months, 1022 MACCEs were recorded, which included 122 deaths, 392 MIs, 156 strokes and 461 instances of TVR. The results from the K‒M analysis suggested that the rates of MACCEs and MI increased with increasing tertiles of the CBR. Furthermore, Cox regression analysis demonstrated that there were significant associations between a high CBR and increased MACCE and MI risk. Conclusion: In subjects who underwent DES implantation, a higher CBR was significantly related to increased long‐term MACCE and MI risk.

Cuff-less methods, especially pulse wave analysis (PWA) techniques with PPG/mmWave sensing, have shown great potential for non-intrusive blood pressure (BP) monitoring. However, the state-of-the-art solutions are only validated on small-scale healthy subjects, neglecting patients with abnormal BP and thus a more urgent need for BP monitoring. To bridge the gap, we first build the largest mmWave-BP dataset to our knowledge, including 930 real patients with cardiovascular diseases, and perform extensive experiments, which reveals that all existing PWA methods exhibit far less satisfactory performance with standard deviation errors (STD) exceeding 16 mmHg for systolic BP (SBP) and 11mmHg for diastolic BP (DBP). An in-depth investigation shows that physiological factors have complex effect on vascular elasticity and structure, thus people with very different BP values may exhibit extremely similar pulse waveform, which leads to confusion in model learning. In this work, we propose BP³, which fuses physiological factors into sensing-data-driven deep-learning framework, so as to capture the intricate effect of physiological factors during the whole process of learning pulse waveforms. Evaluation results show that BP³ achieves the mean errors of-1.57 mmHg and -0.34 mmHg, STD of 9.77 mmHg and 7.93 mmHg for SBP and DBP, respectively. Moreover importantly, BP³ shows remarkable gain particularly for subjects with abnormal BP, achieving mean errors that are only 0.48% ~ 20.86% of the state-of-the-art solutions.