Deep representation learning from electronic medical records identifies distinct symptom based subtypes and progression patterns for COVID-19 prognosis.

OBJECTIVE: Symptoms are significant kind of phenotypes for managing and controlling of the burst of acute infectious diseases, such as COVID-19. Although patterns of symptom clusters and time series have been considered the high potential prediction factors for the prognosis of patients, the elaborated subtypes and their progression patterns based on symptom phenotypes related to the prognosis of COVID-19 patients still need be detected. This study aims to investigate patient subtypes and their progression patterns with distinct features of outcome and prognosis. METHODS: This study included a total of 14,139 longitudinal electronic medical records (EMRs) obtained from four hospitals in Hubei Province, China, involving 2,683 individuals in the early stage of COVID-19 pandemic. A deep representation learning model was developed to help acquire the symptom profiles of patients. K-means clustering algorithm is used to divide them into distinct subtypes. Subsequently, symptom progression patterns were identified by considering the subtypes associated with patients upon admission and discharge. Furthermore, we used Fisher's test to identify significant clinical entities for each subtype. RESULTS: Three distinct patient subtypes exhibiting specific symptoms and prognosis have been identified. Particularly, Subtype 0 includes 44.2% of the whole and is characterized by poor appetite, fatigue and sleep disorders; Subtype 1 includes 25.6% cases and is characterized by confusion, cough with bloody sputum, encopresis and urinary incontinence; Subtype 2 includes 30.2% cases and is characterized by dry cough and rhinorrhea. These three subtypes demonstrate significant disparities in prognosis, with the mortality rates of 4.72%, 8.59%, and 0.25% respectively. Furthermore, symptom cluster progression patterns showed that patients with Subtype 0 who manifest dark yellow urine, chest pain, etc. in the admission stage exhibit an elevated risk of transforming into the more severe subtypes with poor outcome, whereas those presenting with nausea and vomiting tend to incline towards entering the milder subtype. CONCLUSION: This study has proposed a clinical meaningful approach by utilizing the deep representation learning and real-world EMR data containing symptom phenotypes to identify the COVID-19 subtypes and their progression patterns. The results would be potentially useful to help improve the precise stratification and management of acute infectious diseases.

Geometric Amplitude Accompanying Local Responses: Spinor Phase Information from the Amplitudes of Spin-Polarized STM Measurements.

Solving the Hamiltonian of a system yields the energy dispersion and eigenstates. The geometric phase of the eigenstates generates many novel effects and potential applications. However, the geometric properties of the energy dispersion go unheeded. Here, we provide geometric insight into energy dispersion and introduce a geometric amplitude, namely, the geometric density of states (GDOS) determined by the Riemann curvature of the constant-energy contour. The geometric amplitude should accompany various local responses, which are generally formulated by the real-space Green's function. Under the stationary phase approximation, the GDOS simplifies the Green's function into its ultimate form. In particular, the amplitude factor embodies the spinor phase information of the eigenstates, favoring the extraction of the spin texture for topological surface states under an in-plane magnetic field through spin-polarized STM measurements. This work opens a new avenue for exploring the geometric properties of electronic structures and excavates the unexplored potential of spin-polarized STM measurements to probe the spinor phase information of eigenstates from their amplitudes.

Minimum Fascia-Tumour Distance for Differentiating Deep Lobe From Superficial Lobe Benign Parotid Tumours: A Retrospective Study and Meta-Analysis.

BACKGROUND: This study was designed to evaluate the diagnostic efficacy of the minimum fascia-tumour distance (MFTD) in distinguishing deep-lobe benign parotid tumours from superficial-lobe tumours through both an original study and a meta-analysis. METHODS: In this study, we performed a retrospective analysis of data from 91 patients who had been diagnosed with benign parotid tumours. The MFTD values were sourced from preoperative ultrasound examinations. The locations of these tumours were confirmed through surgical findings. We assessed the diagnostic accuracy of MFTD by utilising receiver operating characteristic (ROC) curves. Additionally, we conducted a systematic review of the pertinent literature and performed a diagnostic meta-analysis to ascertain the overall diagnostic efficacy of MFTD in identifying benign parotid tumours. RESULTS: Patients with tumours in the deep lobe had a significantly greater MFTD than patients with tumours in the superficial lobe. Using a cutoff value of 3.50 mm for MFTD, we found an AUC of 0.93, a sensitivity of 81.8%, and a specificity of 98.8%. Our meta-analysis included seven studies covering a total of 1689 tumours. The pooled values for sensitivity, specificity, and diagnostic odds ratio (OR) of MFTD were 81.0%, 89.0%, and 32.2, respectively. The AUC of the summarised ROC curve of MFTD was 0.90. CONCLUSION: The MFTD demonstrated reliable diagnostic accuracy in identifying deep-lobe benign parotid tumours and may be incorporated into standard evaluations before parotidectomy.

Noncanonical formation of SNX5 gene-derived circular RNA regulates cancer growth.

Oral squamous cell carcinoma (OSCC) is a prevalent cancer worldwide, exhibiting unique regional prevalence. Despite advancements in diagnostics and therapy, the 5-year survival rate for patients has seen limited improvement. A deeper understanding of OSCC pathogenesis, especially its molecular underpinnings, is essential for improving detection, prevention, and treatment. In this context, noncoding RNAs, such as circular RNAs (circRNAs), have gained recognition as crucial regulators and potential biomarkers in OSCC progression. Our study highlights the discovery of previously uncharacterized circRNAs, including a SNX5 gene-derived circRNA, circSNX5, through deep sequencing of OSCC patient tissue transcriptomes. We established circSNX5's tumor-specific expression and its strong correlation with patient survival using structure-specific and quantitative PCR analyses. In vitro and in vivo experiments underscored circSNX5 RNA's regulatory role in cancer growth and metastasis. Further, our omics profiling and functional assays revealed that ADAM10 is a critical effector in circSNX5-mediated cancer progression, with circSNX5 maintaining ADAM10 expression by sponging miR-323. This novel circRNA-miRNA-mRNA regulatory axis significantly contributes to oral cancer progression and malignancy. Moreover, we discovered that circSNX5 RNA is produced via noncanonical sequential back-splicing of pre-mRNA, a process negatively regulated by the RNA-binding protein STAU1. This finding adds a new dimension to our understanding of exonic circRNA biogenesis in the eukaryotic transcriptome. Collectively, our findings offer a detailed mechanistic dissection and functional interpretation of a novel circRNA, shedding light on the role of the noncoding transcriptome in cancer biology and potentially paving the way for innovative therapeutic strategies.

Achieving Ultrasound-Excited Emission with Organic Mechanoluminescent Materials.

Unlike traditional photoluminescence (PL), mechanoluminescence (ML) achieved under mechanical excitation demonstrates unique characteristics such as high penetrability, spatial resolution, and signal-to-background ratio (SBR) for bioimaging applications. However, bioimaging with organic mechanoluminescent materials remains challenging because of the shallow penetration depth of ML with short emission wavelengths and the absence of a suitable mechanical force to generate ML in vivo. To resolve these issues, the present paper reports the achievement of ultrasound (US)-excited fluorescence and phosphorescence from purely organic luminogens for the first time with emission wavelengths extending to the red/NIR region, with the penetrability of the US-excited emission being considerably higher than that of PL. Consequently, US-excited subcutaneous phosphorescence imaging can be achieved using a mechanoluminescent-luminogen-based capsule device with a quantified intensity of 9.15 ± 1.32 × 104 p s-1 cm-2 sr-1 and an SBR of 24. Moreover, the US-excited emission can be adequately tuned using the packing modes of the conjugated skeletons, dipole orientation of mechanoluminescent luminogens, and strength and direction of intermolecular interactions. Overall, this study innovatively expands the kind of excitation sources and the emission wavelengths of organic mechanoluminescent materials, paving the way for practical biological applications based on US-excited emission.

Very-large-scale-integrated high quality factor nanoantenna pixels.

Metasurfaces precisely control the amplitude, polarization and phase of light, with applications spanning imaging, sensing, modulation and computing. Three crucial performance metrics of metasurfaces and their constituent resonators are the quality factor (Q factor), mode volume (Vm) and ability to control far-field radiation. Often, resonators face a trade-off between these parameters: a reduction in Vm leads to an equivalent reduction in Q, albeit with more control over radiation. Here we demonstrate that this perceived compromise is not inevitable: high quality factor, subwavelength Vm and controlled dipole-like radiation can be achieved simultaneously. We design high quality factor, very-large-scale-integrated silicon nanoantenna pixels (VINPix) that combine guided mode resonance waveguides with photonic crystal cavities. With optimized nanoantennas, we achieve Q factors exceeding 1,500 with Vm less than 0.1 ( λ / n air ) 3 . Each nanoantenna is individually addressable by free-space light and exhibits dipole-like scattering to the far-field. Resonator densities exceeding a million nanoantennas per cm2 can be achieved. As a proof-of-concept application, we show spectrometer-free, spatially localized, refractive-index sensing, and fabrication of an 8 mm × 8 mm VINPix array. Our platform provides a foundation for compact, densely multiplexed devices such as spatial light modulators, computational spectrometers and in situ environmental sensors.

Defect-Free Nanowelding of Bilayer SnSe Nanoplates.

Nanowelding is a bottom-up technique to create custom-designed nanostructures and devices beyond the precision of lithographic methods. Here, a new technique is reported based on anisotropic lubricity at the van der Waals interface between monolayer and bilayer SnSe nanoplates and a graphene substrate to achieve precise control of the crystal orientation and the interface during the welding process. As-grown SnSe monolayer and bilayer nanoplates are commensurate with graphene's armchair direction but lack commensuration along graphene's zigzag direction, resulting in a reduced friction along that direction and a rail-like, 1D movement that permits joining nanoplates with high precision. This way, molecular beam epitaxially grown SnSe nanoplates of lateral sizes 30-100 nm are manipulated by the tip of a scanning tunneling microscope at room temperature. In situ annealing is applied afterward to weld contacting nanoplates without atomic defects at the interface. This technique can be generalized to any van der Waals interfaces with anisotropic lubricity and is highly promising for the construction of complex quantum devices, such as field effect transistors, quantum interference devices, lateral tunneling junctions, and solid-state qubits.

Nuclear factors NF-YC3 and NF-YBs positively regulate arbuscular mycorrhizal symbiosis in tomato.

The involvement of nuclear factor Y (NF-Y) in transcriptional reprogramming during arbuscular mycorrhizal symbiosis has been demonstrated in several plant species. However, a comprehensive picture is lacking. We showed that the spatial expression of NF-YC3 was observed in cortical cells containing arbuscules via the cis-regulatory element GCC boxes. Moreover, the NF-YC3 promoter was transactivated by the combination of CYCLOPS and autoactive calcium and calmodulin-dependent kinase (CCaMK) via GCC boxes. Knockdown of NF-YC3 significantly reduced the abundance of all intraradical fungal structures and affected arbuscule size. BCP1, SbtM1, and WRI5a, whose expression associated with NF-YC3 levels, might be downstream of NF-YC3. NF-YC3 interacted with NF-YB3a, NF-YB5c, or NF-YB3b, in yeast (Saccharomyces cerevisiae) and in planta, and interacted with NF-YA3a in yeast. Spatial expression of three NF-YBs was observed in all cell layers of roots under both mock and mycorrhizal conditions. Simultaneous knockdown of three NF-YBs, but not individually, reduced the fungal colonization level, suggesting that there might be functional redundancy of NF-YBs to regulate AM symbiosis. Collectively, our data suggest that NF-YC3 and NF-YBs positively regulate AM symbiosis in tomato, and arbuscule-related NF-YC3 may be an important downstream gene of the common symbiosis signaling pathway.

Prospective Impacts of Social Participation Activities on Cognitive and Psychopathological Statuses in Community-Dwelling Older Adults.

OBJECTIVES: This study investigated the impact of social activities on cognitive functioning and psychopathological symptoms. METHODS: Participants aged 55 or older were enrolled through communities. Initial measures assessed demographic data, neuropsychological functioning, psychopathological state, and happiness. Social activities were evaluated using a modified 12-item tool, with 3-4 activities as the cutoff. Follow-up after 6-9 months included Mini-Mental State Examination (MMSE), Beck Depression Inventory - II (BDI-II), Beck Anxiety Inventory (BAI), Health Assessment Questionnaire (HAQ), and Patient Health Questionnaire-15 (PHQ-15) measurements. Predictive models for psychiatric and cognitive statuses were built using multiple linear regression, adjusting for baseline conditions. RESULTS: Initially, 516 older individuals enrolled, with 403 undergoing follow-up. During follow-up, the low participation group reported lower MMSE scores, higher BAI scores, and increased PHQ-15 risk. Negative correlations between social activity numbers and PHQ-15 results were found. Engagement in social clubs correlated positively with higher MMSE scores, while regular interactions with one's adult child(ren) were linked to decreased BAI scores. CONCLUSIONS: The quantity of social activities was associated with lower somatic distress. Social club engagement positively influenced cognition, and regular interactions with one's adult child(ren) mitigated anxiety among older individuals. CLINICAL IMPLICATIONS: Enough types of social activities, participating in social clubs, and adequate interactions with children protected against psychopathologies.

From Genotype to Phenotype: Raman Spectroscopy and Machine Learning for Label-Free Single-Cell Analysis.

Raman spectroscopy has made significant progress in biosensing and clinical research. Here, we describe how surface-enhanced Raman spectroscopy (SERS) assisted with machine learning (ML) can expand its capabilities to enable interpretable insights into the transcriptome, proteome, and metabolome at the single-cell level. We first review how advances in nanophotonics-including plasmonics, metamaterials, and metasurfaces-enhance Raman scattering for rapid, strong label-free spectroscopy. We then discuss ML approaches for precise and interpretable spectral analysis, including neural networks, perturbation and gradient algorithms, and transfer learning. We provide illustrative examples of single-cell Raman phenotyping using nanophotonics and ML, including bacterial antibiotic susceptibility predictions, stem cell expression profiles, cancer diagnostics, and immunotherapy efficacy and toxicity predictions. Lastly, we discuss exciting prospects for the future of single-cell Raman spectroscopy, including Raman instrumentation, self-driving laboratories, Raman data banks, and machine learning for uncovering biological insights.

Development of a tool measuring various aspects of social detachment: The social detachment questionnaire for older population.

BACKGROUND: Social detachment includes the subjective aspect "loneliness" and the objective aspect "social isolation," but tools to assess both dimensions are limited. This study aims to develop a questionnaire, the Social Detachment Questionnaire for Older Population (SDQO), that considers multiple dimensions of social detachment simultaneously. METHODS: The study collected 600 valid samples from individuals aged 55 and above to examine the psychometric properties of the developed SDQO. Item analysis was conducted to assess the performance of each item, and exploratory factor analysis (EFA) was employed to analyze its initial structure and eliminate less ideal items. Subsequently, confirmatory factor analysis (CFA) was used to examine the model fit of the suggested structure by EFA, using different subsamples. Internal consistency, concurrent validity, and other analyses were also performed. RESULTS: The original 27-item SDQO was reduced to 17 items after removing 4 questions in item analysis and 6 questions in EFA. The Cronbach's alpha for the 17-item version of SDQO was 0.80. Both EFA and CFA supported its 6-factor structure, with factors identified as community activities, loneliness, personal resources, leisure activities, friendship, and family resources. SDQO also demonstrated expected performance in concurrent validity. CONCLUSION: The 17-item version of SDQO exhibited good reliability and validity, measuring various aspects of social detachment behavior, feelings, and resources. It holds value for future research applications.

Gut Bifidobacterium longum is associated with better native liver survival in patients with biliary atresia.

Background & Aims: The gut microbiome plays an important role in liver diseases, but its specific impact on biliary atresia (BA) remains to be explored. We aimed to investigate the microbial signature in the early life of patients with BA and to analyze its influence on long-term outcomes. Methods: Fecal samples (n = 42) were collected from infants with BA before and after Kasai portoenterostomy (KPE). The stool microbiota was analyzed using 16S rRNA next-generation sequencing and compared with that of age-matched healthy controls (HCs). Shotgun metagenomic sequencing analysis was employed to confirm the bacterial composition in 10 fecal samples before KPE. The correlation of the microbiome signature with liver function and long-term outcomes was assessed. Results: In the 16S rRNA next-generation sequencing analysis of fecal microbiota, the alpha and beta diversity analyses revealed significant differences between HCs and patients with BA before and after KPE. The difference in microbial composition analyzed by linear discriminant analysis and random forest classification revealed that the abundance of Bifidobacterium longum (B. longum) was significantly lower in patients before and after KPE than in HCs. The abundance of B. longum was negatively correlated with the gamma-glutamyltransferase level after KPE (p <0.05). Patients with early detectable B. longum had significantly lower total and direct bilirubin 3 months after KPE (p <0.005) and had a significantly lower liver transplantation rate (hazard ratio: 0.16, 95% CI 0.03-0.83, p = 0.029). Shotgun metagenomic sequencing also revealed that patients with BA and detectable B. longum had reduced total and direct bilirubin after KPE. Conclusion: The gut microbiome of patients with BA differed from that of HCs, with a notable abundance of B. longum in early infancy correlating with better long-term outcomes. Impact and implications: Bifidobacterium longum (B. longum) is a beneficial bacterium commonly found in the human gut. It has been studied for its potential impacts on various health conditions. In patients with biliary atresia, we found that a greater abundance of B. longum in the fecal microbiome is associated with improved clinical outcomes. This suggests that early colonization and increasing B. longum levels in the gut could be a therapeutic strategy to improve the prognosis of patients with biliary atresia.

Medical Vision-Language Pre-Training for Brain Abnormalities.

Vision-language models have become increasingly powerful for tasks that require an understanding of both visual and linguistic elements, bridging the gap between these modalities. In the context of multimodal clinical AI, there is a growing need for models that possess domain-specific knowledge, as existing models often lack the expertise required for medical applications. In this paper, we take brain abnormalities as an example to demonstrate how to automatically collect medical image-text aligned data for pretraining from public resources such as PubMed. In particular, we present a pipeline that streamlines the pre-training process by initially collecting a large brain image-text dataset from case reports and published journals and subsequently constructing a high-performance vision-language model tailored to specific medical tasks. We also investigate the unique challenge of mapping subfigures to subcaptions in the medical domain. We evaluated the resulting model with quantitative and qualitative intrinsic evaluations. The resulting dataset and our code can be found here https://github.com/masoud-monajati/MedVL_pretraining_pipeline.

Limbic Network and Papez Circuit Involvement in ALS: Imaging and Clinical Profiles in GGGGCC Hexanucleotide Carriers in C9orf72 and C9orf72-Negative Patients.

Background: While frontotemporal involvement is increasingly recognized in Amyotrophic lateral sclerosis (ALS), the degeneration of limbic networks remains poorly characterized, despite growing evidence of amnestic deficits, impaired emotional processing and deficits in social cognition. Methods: A prospective neuroimaging study was conducted with 204 individuals with ALS and 111 healthy controls. Patients were stratified for hexanucleotide expansion status in C9orf72. A deep-learning-based segmentation approach was implemented to segment the nucleus accumbens, hypothalamus, fornix, mammillary body, basal forebrain and septal nuclei. The cortical, subcortical and white matter components of the Papez circuit were also systematically evaluated. Results: Hexanucleotide repeat expansion carriers exhibited bilateral amygdala, hypothalamus and nucleus accumbens atrophy, and C9orf72 negative patients showed bilateral basal forebrain volume reductions compared to controls. Both patient groups showed left rostral anterior cingulate atrophy, left entorhinal cortex thinning and cingulum and fornix alterations, irrespective of the genotype. Fornix, cingulum, posterior cingulate, nucleus accumbens, amygdala and hypothalamus degeneration was more marked in C9orf72-positive ALS patients. Conclusions: Our results highlighted that mesial temporal and parasagittal subcortical degeneration is not unique to C9orf72 carriers. Our radiological findings were consistent with neuropsychological observations and highlighted the importance of comprehensive neuropsychological testing in ALS, irrespective of the underlying genotype.

IoT Privacy Risks Revealed.

A user's devices such as their phone and computer are constantly bombarded by IoT devices and associated applications seeking connection to the user's devices. These IoT devices may or may not seek explicit user consent, thus leaving the users completely unaware the IoT device is collecting, using, and/or sharing their personal data or, only marginal informed, if the user consented to the connecting IoT device but did not read the associated privacy policies. Privacy policies are intended to inform users of what personally identifiable information (PII) data will be collected about them and the policies about how those PII data will be used and shared. This paper presents novel tools and the underlying algorithms employed by the Personal Privacy Assistant app (UTCID PPA) developed by the University of Texas at Austin Center for Identity to inform users of IoT devices seeking to connect to their devices and to notify those users of potential privacy risks posed by the respective IoT device. The assessment of these privacy risks must deal with the uncertainty associated with sharing the user's personal data. If privacy risk (R) equals the consequences (C) of an incident (i.e., personal data exposure) multiplied by the probability (P) of those consequences occurring (C × P), then efforts to control risks must seek to reduce the possible consequences of an incident as well as reduce the uncertainty of the incident and its consequences occurring. This research classifies risk according to two parameters: expected value of the incident's consequences and uncertainty (entropy) of those consequences. This research calculates the entropy of the privacy incident consequences by evaluating: (1) the data sharing policies governing the IoT resource and (2) the type of personal data exposed. The data sharing policies of an IoT resource are scored by the UTCID PrivacyCheck™, which uses machine learning to read and score the IoT resource privacy policies against metrics set forth by best practices and international regulations. The UTCID Identity Ecosystem uses empirical identity theft and fraud cases to assess the entropy of privacy incident consequences involving a specific type of personal data, such as name, address, Social Security number, fingerprint, and user location. By understanding the entropy of a privacy incident posed by a given IoT resource seeking to connect to a user's device, UTCID PPA offers actionable recommendations enhancing the user's control over IoT connections, interactions, their personal data, and, ultimately, user-centric privacy control.

Label-free photoacoustic computed tomography of visually evoked responses in the primary visual cortex and four subcortical retinorecipient nuclei of anesthetized mice.

Significance: Many techniques exist for screening retinal phenotypes in mouse models in vision research, but significant challenges remain for efficiently probing higher visual centers of the brain. Photoacoustic computed tomography (PACT), with optical sensitivity to hemodynamic response (HR) in brain and ultrasound resolution, provides unique advantages in comprehensively assessing higher visual function in the mouse brain. Aim: We aim to examine the reliability of PACT in the functional phenotyping of mouse models for vision research. Approach: A PACT-ultrasound (US) parallel imaging system was established with a one-dimensional (1D) US transducer array and a tunable laser. Imaging was performed at three coronal planes of the brain, covering the primary visual cortex and the four subcortical nuclei, including the superior colliculus, the dorsal lateral geniculate nucleus, the suprachiasmatic nucleus, and the olivary pretectal nucleus. The visual-evoked HR was isolated from background signals using an impulse-based data processing protocol. rd1 mice with rod/cone degeneration, melanopsin-knockout (mel-KO) mice with photoreceptive ganglion cells that lack intrinsic photosensitivity, and wild-type mice as controls were imaged. The quantitative characteristics of the visual-evoked HR were compared. Results: Quantitative analysis of the HRs shows significant differences among the three mouse strains: (1) rd1 mice showed both smaller and slower responses compared with wild type ( n = 10,10 , p < 0.01 ) and (2) mel-KO mice had lower amplitude but not significantly delayed photoresponses than wild-type mice ( n = 10,10 , p < 0.01 ). These results agree with the known visual deficits of the mouse strains. Conclusions: PACT demonstrated sufficient sensitivity to detecting post-retinal functional deficits.

Effect of body mass index on mortality for diabetic patients with aortic stenosis.

BACKGROUND: Several studies suggest an "obesity paradox," associating obesity with better cardiovascular outcomes in patients with type 2 diabetes mellitus (DM) or aortic stenosis (AS) compared to normal or underweight individuals. This study explores the impact of body mass index (BMI) on diabetic patients with AS. METHODS: Between 2014 and 2019, patients with DM who underwent echocardiography were analyzed. Outcomes included all-cause mortality, cardiovascular, and non-cardiovascular death. Patients were categorized as underweight, normal weight, or obese based on BMI (<18.5, 18.5 to 27, and >27 kg/m2, respectively). RESULTS: Among 74,835 DM patients, 734 had AS. Normal weight comprised 65.5% (n=481), underweight 4.1% (N=30), and 30.4% were obese. Over a 6-year follow-up, underweight patients had significantly higher all-cause mortality (HR 1.96, 95% CI 1.22 - 3.14, p = 0.005), while obese patients had significantly lower mortality (HR 0.79, 95% CI 0.68 - 0.91, p=0.001) compared to the normal group. Regarding etiologies, underweight patients had a higher risk of non-cardiovascular death (HR 2.47, 95% CI 1.44-4.25, p = 0.001), while obese patients had a lower risk of cardiovascular death (HR 0.66, 95% CI 0.50-0.86, p=0.003). Subgroup analysis showed a consistent trend without significant interaction. CONCLUSIONS: BMI significantly impacts mortality in DM patients with AS. Being underweight is associated with worse non-cardiovascular death, while obesity is linked to improved cardiovascular death outcomes.

Lingdan: enhancing encoding of traditional Chinese medicine knowledge for clinical reasoning tasks with large language models.

OBJECTIVE: The recent surge in large language models (LLMs) across various fields has yet to be fully realized in traditional Chinese medicine (TCM). This study aims to bridge this gap by developing a large language model tailored to TCM knowledge, enhancing its performance and accuracy in clinical reasoning tasks such as diagnosis, treatment, and prescription recommendations. MATERIALS AND METHODS: This study harnessed a wide array of TCM data resources, including TCM ancient books, textbooks, and clinical data, to create 3 key datasets: the TCM Pre-trained Dataset, the Traditional Chinese Patent Medicine (TCPM) Question Answering Dataset, and the Spleen and Stomach Herbal Prescription Recommendation Dataset. These datasets underpinned the development of the Lingdan Pre-trained LLM and 2 specialized models: the Lingdan-TCPM-Chat Model, which uses a Chain-of-Thought process for symptom analysis and TCPM recommendation, and a Lingdan Prescription Recommendation model (Lingdan-PR) that proposes herbal prescriptions based on electronic medical records. RESULTS: The Lingdan-TCPM-Chat and the Lingdan-PR Model, fine-tuned on the Lingdan Pre-trained LLM, demonstrated state-of-the art performances for the tasks of TCM clinical knowledge answering and herbal prescription recommendation. Notably, Lingdan-PR outperformed all state-of-the-art baseline models, achieving an improvement of 18.39% in the Top@20 F1-score compared with the best baseline. CONCLUSION: This study marks a pivotal step in merging advanced LLMs with TCM, showcasing the potential of artificial intelligence to help improve clinical decision-making of medical diagnostics and treatment strategies. The success of the Lingdan Pre-trained LLM and its derivative models, Lingdan-TCPM-Chat and Lingdan-PR, not only revolutionizes TCM practices but also opens new avenues for the application of artificial intelligence in other specialized medical fields. Our project is available at https://github.com/TCMAI-BJTU/LingdanLLM.

Clinical experience of infantile hepatic hemangioma.

PURPOSE: To describe the clinical presentation, treatment preference, and relevant complications of infantile hepatic hemangioma (IHH) in propranolol era. METHODS: The National Taiwan University Hospital integrated Medical Database (NTUH-iMD) was used to enroll twenty-one cases of IHH diagnosed from 2006 to 2020. Medical charts were retrospectively reviewed. RESULTS: In nine patients (42.9%), IHH was found incidentally, and in seven patients (33%), it was detected during postnatal self-paid ultrasonography. Focal disease was determined in 17 patients, multifocal disease in 1 patient, and diffuse disease in 3 patients. Patients with diffuse disease had a lower hemoglobulin level than patients with focal IHH (9.38 vs. 12.6 mg/dL, p = 0.045). Two patients had Kasabach-Merritt phenomenon (KMP), one had hypothyroidism, and one had both. All patients with KMP had focal hepatic hemangiomas. Among the 17 patients with focal IHH, nine were prescribed propranolol, one was treated by surgical resection of the tumor, and the others had expectant management. All patients with multifocal and diffuse IHH were administered propranolol. One infant (7.7%) treated with propranolol had bradycardia initially but it subsided after dose adjustment. CONCLUSIONS: Most IHH is found incidentally or detected during postnatal ultrasonography screening. Patients with large focal lesions should also be screened for associated complications. Propranolol is the drug of choice and a safe therapeutic option for IHH, especially for focal tumors >5 cm as well as multifocal and diffuse lesions.