Search

BackgroundUnfractionated heparin is widely used in the intensive care unit as an anticoagulant. However, weight-based heparin dosing has been shown to be suboptimal and may place patients at unnecessary risk during their intensive care unit stay. ObjectiveIn this study, we intended to develop and validate a machine learning–based model to predict heparin treatment outcomes and to provide dosage recommendations to clinicians. MethodsA shallow neural network model was adopted in a retrospective cohort of patients from the Multiparameter Intelligent Monitoring in Intensive Care III (MIMIC III) database and patients admitted to the Peking Union Medical College Hospital (PUMCH). We modeled the subtherapeutic, normal, and supratherapeutic activated partial thromboplastin time (aPTT) as the outcomes of heparin treatment and used a group of clinical features for modeling. Our model classifies patients into 3 different therapeutic states. We tested the prediction ability of our model and evaluated its performance by using accuracy, the kappa coefficient, precision, recall, and the F1 score. Furthermore, a dosage recommendation module was designed and evaluated for clinical decision support. ResultsA total of 3607 patients selected from MIMIC III and 1549 patients admitted to the PUMCH who met our criteria were included in this study. The shallow neural network model showed results of F1 scores 0.887 (MIMIC III) and 0.925 (PUMCH). When compared with the actual dosage prescribed, our model recommended increasing the dosage for 72.2% (MIMIC III, 1240/1718) and 64.7% (PUMCH, 281/434) of the subtherapeutic patients and decreasing the dosage for 80.9% (MIMIC III, 504/623) and 76.7% (PUMCH, 277/361) of the supratherapeutic patients, suggesting that the recommendations can contribute to clinical improvements and that they may effectively reduce the time to optimal dosage in the clinical setting. ConclusionsThe evaluation of our model for predicting heparin treatment outcomes demonstrated that the developed model is potentially applicable for reducing the misdosage of heparin and for providing appropriate decision recommendations to clinicians.

BackgroundCOVID-19 cases resurged worldwide in the second half of 2020. Not much is known about the changes in public responses to containment measures from the initial outbreak to resurgence. Monitoring public responses is crucial to inform policy measures to prepare for COVID-19 resurgence. ObjectiveThis study aimed to assess and compare public responses to containment measures during the initial outbreak and resurgence of COVID-19 in China. MethodsWe curated all COVID-19–related posts from Sina Weibo (China’s version of Twitter) during the initial outbreak and resurgence of COVID-19 in Beijing, China. With a Python script, we constructed subsets of Weibo posts focusing on 3 containment measures: lockdown, the test-trace-isolate strategy, and suspension of gatherings. The Baidu open-source sentiment analysis model and latent Dirichlet allocation topic modeling, a widely used machine learning algorithm, were used to assess public engagement, sentiments, and frequently discussed topics on each containment measure. ResultsA total of 8,985,221 Weibo posts were curated. In China, the containment measures evolved from a complete lockdown for the general population during the initial outbreak to a more targeted response strategy for high-risk populations during COVID-19 resurgence. Between the initial outbreak and resurgence, the average daily proportion of Weibo posts with negative sentiments decreased from 57% to 47% for the lockdown, 56% to 51% for the test-trace-isolate strategy, and 55% to 48% for the suspension of gatherings. Among the top 3 frequently discussed topics on lockdown measures, discussions on containment measures accounted for approximately 32% in both periods, but those on the second-most frequently discussed topic shifted from the expression of negative emotions (11%) to its impacts on daily life or work (26%). The public expressed a high level of panic (21%) during the initial outbreak but almost no panic (1%) during resurgence. The more targeted test-trace-isolate measure received the most support (60%) among all 3 containment measures in the initial outbreak, and its support rate approached 90% during resurgence. ConclusionsCompared to the initial outbreak, the public expressed less engagement and less negative sentiments on containment measures and were more supportive toward containment measures during resurgence. Targeted test-trace-isolate strategies were more acceptable to the public. Our results indicate that when COVID-19 resurges, more targeted test-trace-isolate strategies for high-risk populations should be promoted to balance pandemic control and its impact on daily life and the economy.

BackgroundCOVID-19 is an emerging infectious disease that has created health care challenges worldwide. Pregnant women are particularly affected by this disease. ObjectiveThe aims of this study are to assess the levels of perceived threat (susceptibility, severity, impact), negative emotions (fear, worry), and self-efficacy of pregnant women in China related to COVID-19 and to examine their associations with mental health (depression and anxiety) and personal protective behavior (wearing a face mask). MethodsA total of 4087 pregnant women from China completed a cross-sectional web-based survey between March 3 and 10, 2020. ResultsThe prevalence of probable depression and anxiety was 48.7% (1989/4087) and 33.0% (1347/4087), respectively; 23.8% participants (974/4087) reported always wearing a face mask when going out. Of the 4087 participants, 32.1% (1313) and 36.4% (1490) perceived themselves or their family members to be susceptible to COVID-19 infection, respectively; 3216-3518 (78.7%-86.1%) agreed the disease would have various severe consequences. Additionally, 2275 of the 4087 participants (55.7%) showed self-efficacy in protecting themselves from contracting COVID-19, and 2232 (54.6%) showed efficacy in protecting their family members; 1303 (31.9%) reported a high level of fear of the disease, and 2780-3056 (68.0%-74.8%) expressed worry about various aspects of COVID-19. The results of the multivariate multinominal logistic regression analyses showed that perceived severity, perceived impact, fear, and worry were risk factors for probable depression and anxiety, while self-efficacy was a protective factor. The results of the multivariate logistic regression analysis showed that perceived susceptibility was associated with always wearing a face mask. ConclusionsChinese pregnant women showed high levels of mental distress but low levels of personal protective behavior during the COVID-19 pandemic. Interventions are needed to promote the mental health and health behavior of pregnant women during the pandemic.

BackgroundAlthough lockdown and mandatory quarantine measures have played crucial roles in the sharp decrease of the number of newly confirmed/suspected COVID-19 cases, concerns have been raised over the threat that these measures pose to mental health, especially the mental health of vulnerable groups, including pregnant women. Few empirical studies have assessed whether and how these control measures may affect mental health, and no study has investigated the prevalence and impacts of the use of eHealth resources among pregnant women during the COVID-19 outbreak. ObjectiveThis study investigated (1) the effects of lockdown and mandatory quarantine on mental health problems (ie, anxiety and depressive symptoms), (2) the potential mediation effects of perceived social support and maladaptive cognition, and (3) the moderation effects of eHealth-related factors (ie, using social media to obtain health information and using prenatal care services during the COVID-19 pandemic) on pregnant women in China. MethodsAn online cross-sectional survey was conducted among 19,515 pregnant women from all 34 Chinese provincial-level administrative regions from February 25 to March 10, 2020. ResultsOf the 19,515 participants, 12,209 (62.6%) were subjected to lockdown in their areas of residence, 737 (3.8%) were subjected to mandatory quarantine, 8712 (44.6%) had probable mild to severe depression, 5696 (29.2%) had probable mild to severe anxiety, and 1442 (7.4%) had suicidal ideations. Only 640 (3.3%) participants reported that they used online prenatal care services during the outbreak. Significant sociodemographic/maternal factors of anxiety/depressive symptoms included age, education, occupation, the area of residence, gestational duration, the number of children born, complication during pregnancy, the means of using prenatal care services, and social media use for obtaining health information. Multiple indicators multiple causes modeling (χ214=495.21; P

BackgroundOsteoarthritis (OA) is a chronic, debilitating, and degenerative joint disease. However, it is difficult for patients with knee OA to access conventional rehabilitation when discharging from the hospital. Internet-based rehabilitation is one of the promising telemedicine strategies to provide a means combining monitoring, guidance, and treatment for patients with knee OA. ObjectiveThe aim of this study was to conduct a systematic review and meta-analysis for assessing the effect of internet-based rehabilitation programs on pain and physical function in patients with knee OA. MethodsKeywords related to knee OA and internet-based rehabilitation were systematically searched in the Web of Science, MEDLINE, EMBASE, CENTRAL, Scopus, PEDro (Physiotherapy Evidence Database), CNKI, SinoMed, and WANFANG databases from January 2000 to April 2020. Only randomized controlled trials were included. The authors independently screened the literature. The main outcome measures were focused on pain and physical function. A meta-analysis was performed on the collected data. Review Manager (RevMan, version 5.3) was used for all analyses. ResultsThe systematic review identified 6 randomized controlled trials, 4 of which were included in the meta-analysis, comprising a total of 791 patients with knee OA. The meta-analysis with the fixed-effects model showed that the internet-based rehabilitation programs could significantly alleviate the osteoarthritic pain for patients compared with conventional rehabilitation (standardized mean difference [SMD] –0.21, 95% CI −0.4 to –0.01, P=.04). No significant difference was found in the improvement of physical function in patients with knee OA compared with conventional rehabilitation within 2 to 12 months (SMD –0.08, 95% CI −0.27 to 0.12, P=.43). ConclusionsThis systematic review shows that internet-based rehabilitation programs could improve the pain but not physical function for patients with knee OA. However, there was a very small number of studies that could be included in the review and meta-analysis. Thus, further studies with large sample sizes are warranted to promote the effectiveness of internet-based rehabilitation and to develop its personalized design.

BackgroundInternet addiction has become a major global concern and a burden on mental health. However, there is a lack of consensus on its link to mental health outcomes. ObjectiveThe aim of this study was to investigate the associations between internet addiction severity and adverse mental health outcomes. MethodsFirst-year undergraduates enrolled at Sichuan University during September 2015, 2016, 2017, and 2018 were invited to participate in the current study survey, 85.13% (31,659/37,187) of whom fully responded. Young’s 20-item Internet Addiction Test, Patient Health Questionnaire-15, Patient Health Questionnaire-9, Symptom Checklist 90, Six-Item Kessler Psychological Distress Scale, and Suicidal Behaviors Questionnaire-Revised were used to evaluate internet addiction, four psychopathologies (high somatic symptom severity, clinically significant depression, psychoticism, and paranoia), serious mental illness, and lifetime suicidality. ResultsThe prevalence of students with mild, moderate, and severe internet addiction was 37.93% (12,009/31,659), 6.33% (2003/31,659), and 0.20% (63/31,659), respectively. The prevalence rates of high somatic symptom severity, clinically significant depression, psychoticism, paranoid ideation, and serious mental illness were 6.54% (2072/31,659), 4.09% (1294/31,659), 0.51% (160/31,659), 0.52% (165/31,659), and 1.88% (594/31,659), respectively, and the lifetime prevalence rates of suicidal ideation, suicidal plan, and suicidal attempt were 36.31% (11,495/31,659), 5.13% (1624/31,659), and 1.00% (315/31,659), respectively. The prevalence rates and odds ratios (ORs) of the four psychopathologies and their comorbidities, screened serious mental illness, and suicidalities in the group without internet addiction were much lower than the average levels of the surveyed population. Most of these metrics in the group with mild internet addiction were similar to or slightly higher than the average rates; however, these rates sharply increased in the moderate and severe internet addiction groups. Among the four psychopathologies, clinically significant depression was most strongly associated with internet addiction after adjusting for the confounding effects of demographics and other psychopathologies, and its prevalence increased from 1.01% (178/17,584) in the students with no addiction to 4.85% (582/12,009), 24.81% (497/2,003), and 58.73% (37/63) in the students with mild, moderate, and severe internet addiction, respectively. The proportions of those with any of the four psychopathologies increased from 4.05% (713/17,584) to 11.72% (1408/12,009), 36.89% (739/2003), and 68.25% (43/63); those with lifetime suicidal ideation increased from 24.92% (4382/17,584) to 47.56% (5711/12,009), 67.70% (1356/2003), and 73.02% (46/63); those with a suicidal plan increased from 2.59% (456/17,584) to 6.77% (813/12,009), 16.72% (335/2003), and 31.75% (20/63); and those with a suicidal attempt increased from 0.50% (88/17,584) to 1.23% (148/12,009), 3.54% (71/2003), and 12.70% (8/63), respectively. ConclusionsModerate and severe internet addiction were strongly associated with a broad group of adverse mental health outcomes, including somatic symptoms that are the core features of many medical illnesses, although clinically significant depression showed the strongest association. This finding supports the illness validity of moderate and severe internet addiction in contrast to mild internet addiction. These results are important for informing health policymakers and service suppliers from the perspective of resolving the overall human health burden in the current era of “Internet Plus” and artificial intelligence.