Longmei Tang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Longmei Tang has a total of 94 co-authors. Here are the top co-authors by number of shared publications:
- Jin Tian
Hospital Management Innovation Research Center, The First Hospital of Hebei Medical University, Shijiazhuang, China
5 shared publications Boyan Li
School of Public Health, Hebei Medical University, Shijiazhuang, China
3 shared publicationsDan Li
School of Public Health, Hebei Medical University, Shijiazhuang, China
3 shared publicationsHuixia Shi
School of Public Health, Hebei Medical University, Shijiazhuang, China
3 shared publicationsJiajun Li
School of Public Health, Hebei Medical University, Shijiazhuang, China
2 shared publicationsJie Liu
Senior Department of Cardiology, The Sixth Medical Center of PLA General Hospital, Beijing, China
2 shared publications
Latest Publications
- Open Access
- 27 September 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Nonlinear joint modeling of oxygenation indices for dynamic mortality prediction in neonates with persistent pulmonary hypertension receiving inhaled nitric oxide
Persistent pulmonary hypertension of the newborn (PPHN) remains a life-threatening disorder with considerable mortality despite the widespread use of inhaled nitric oxide (iNO). Current risk assessment relies on static indices and cannot capture the rapid physiological changes characteristic of this disease. We aimed to evaluate the association between dynamic oxygenation indices and mortality, and to develop a nonlinear joint modeling framework for individualized prediction in neonates receiving iNO therapy. In this retrospective cohort of 377 PPHN neonates, serial blood gas and ventilator data were used to compute oxygen saturation index (OSI) and oxygen index (OI). Linear and nonlinear joint models linking longitudinal OSI/OI trajectories with time-to-death were constructed via Bayesian estimation. Results Non-survivors showed persistently higher OSI and OI trajectories than survivors (p < 0.001). Among six candidate models, the nonlinear OSI-based joint model achieved the best performance (lowest DIC/WAIC, highest LPML) and excellent discrimination (24-h AUC = 0.93, cross-validation AUC = 0.90, validation AUC = 0.84), substantially outperforming linear (validation AUC: 0.78–0.82), OI-based (validation AUC: 0.75–0.83), time-varying Cox (validation AUC: 0.74–0.79), and nSOFA-based models (validation AUC: 0.68–0.70). Dynamic predictions demonstrated that early OSI trajectories strongly influenced survival probability. Dynamic OSI trajectories are independent predictors of mortality in PPHN neonates. Nonlinear OSI-based joint modeling enables individualized, time-updated mortality prediction within the first 24 h of iNO therapy and may inform early clinical decision-making.
- 27 September 2026
- Published by Springer Nature in Scientific Reports
- Vol. 16 (1)
Open AccessNonlinear joint modeling of oxygenation indices for dynamic mortality prediction in neonates with persistent pulmonary hypertension receiving inhaled nitric oxide
Persistent pulmonary hypertension of the newborn (PPHN) remains a life-threatening disorder with considerable mortality despite the widespread use of inhaled nitric oxide (iNO). Current risk assessment relies on static indices and cannot capture the rapid physiological changes characteristic of this disease. We aimed to evaluate the association between dynamic oxygenation indices and mortality, and to develop a nonlinear joint modeling framework for individualized prediction in neonates receiving iNO therapy. In this retrospective cohort of 377 PPHN neonates, serial blood gas and ventilator data were used to compute oxygen saturation index (OSI) and oxygen index (OI). Linear and nonlinear joint models linking longitudinal OSI/OI trajectories with time-to-death were constructed via Bayesian estimation. Results Non-survivors showed persistently higher OSI and OI trajectories than survivors (p < 0.001). Among six candidate models, the nonlinear OSI-based joint model achieved the best performance (lowest DIC/WAIC, highest LPML) and excellent discrimination (24-h AUC = 0.93, cross-validation AUC = 0.90, validation AUC = 0.84), substantially outperforming linear (validation AUC: 0.78–0.82), OI-based (validation AUC: 0.75–0.83), time-varying Cox (validation AUC: 0.74–0.79), and nSOFA-based models (validation AUC: 0.68–0.70). Dynamic predictions demonstrated that early OSI trajectories strongly influenced survival probability. Dynamic OSI trajectories are independent predictors of mortality in PPHN neonates. Nonlinear OSI-based joint modeling enables individualized, time-updated mortality prediction within the first 24 h of iNO therapy and may inform early clinical decision-making.
- Open Access
- 17 September 2026
- Published by Springer Nature in BMC Medical Education
Evaluating the conditions for practice-based learning: an expert-informed Input–Process–Output framework for Master of Public Health programmes in China
Background: Preparing Master of Public Health (MPH) graduates for practice requires opportunities to apply evidence to population-health problems and receive feedback on their work. This study developed an expert-informed Input–Process–Output (IPO) framework for evaluating how Chinese MPH programmes organise practical training, supervision and assessment within the wider programme. Methods: Policy and institutional documents, published research and interviews with 20 public health professionals in Hebei Province informed 49 candidate indicators. Two Delphi rounds, involving 20 and 18 experts, assessed importance, scientific soundness and feasibility. Item ratings were summarised descriptively, and tie-corrected Kendall’s W assessed concordance in relative rankings. Fuzzy analytic hierarchy weighting used the first author’s pairwise judgements, with calculations jointly checked by the first and corresponding authors. Products of local weights were normalised to obtain global indicator weights. Results: The final framework comprised three primary dimensions, nine secondary domains, 19 tertiary subdomains and 45 fourth-level indicators. Round 2 mean ratings ranged from 3.89 to 4.67 across the three rating dimensions, with all coefficients of variation ≤ 0.25. Rank concordance was low: Kendall’s W was 0.0529 for importance, 0.0569 for scientific soundness and 0.0474 for feasibility (all P > 0.05). Process received the highest primary weight (0.481), followed by Input (0.266) and Output (0.253). Specification of practice-mentor guidance time had the highest normalised global weight (0.080508), followed by dual-qualified faculty and four equally weighted indicators concerning practical training and assessment. Conclusions: The framework identifies candidate conditions for practice-based MPH learning and offers a structure for reviewing practical training, supervision and assessment. Empirical evaluation is needed to determine whether these arrangements are implemented as intended and associated with students’ public health performance.
- 17 September 2026
- Published by Springer Nature in BMC Medical Education
Open AccessEvaluating the conditions for practice-based learning: an expert-informed Input–Process–Output framework for Master of Public Health programmes in China
Background: Preparing Master of Public Health (MPH) graduates for practice requires opportunities to apply evidence to population-health problems and receive feedback on their work. This study developed an expert-informed Input–Process–Output (IPO) framework for evaluating how Chinese MPH programmes organise practical training, supervision and assessment within the wider programme. Methods: Policy and institutional documents, published research and interviews with 20 public health professionals in Hebei Province informed 49 candidate indicators. Two Delphi rounds, involving 20 and 18 experts, assessed importance, scientific soundness and feasibility. Item ratings were summarised descriptively, and tie-corrected Kendall’s W assessed concordance in relative rankings. Fuzzy analytic hierarchy weighting used the first author’s pairwise judgements, with calculations jointly checked by the first and corresponding authors. Products of local weights were normalised to obtain global indicator weights. Results: The final framework comprised three primary dimensions, nine secondary domains, 19 tertiary subdomains and 45 fourth-level indicators. Round 2 mean ratings ranged from 3.89 to 4.67 across the three rating dimensions, with all coefficients of variation ≤ 0.25. Rank concordance was low: Kendall’s W was 0.0529 for importance, 0.0569 for scientific soundness and 0.0474 for feasibility (all P > 0.05). Process received the highest primary weight (0.481), followed by Input (0.266) and Output (0.253). Specification of practice-mentor guidance time had the highest normalised global weight (0.080508), followed by dual-qualified faculty and four equally weighted indicators concerning practical training and assessment. Conclusions: The framework identifies candidate conditions for practice-based MPH learning and offers a structure for reviewing practical training, supervision and assessment. Empirical evaluation is needed to determine whether these arrangements are implemented as intended and associated with students’ public health performance.
- Open Access
- 14 August 2026
- Published by Wolters Kluwer Health in Medicine
- Vol. 105 (33) , e50080
Association between air pollution and acute exacerbation of chronic obstructive pulmonary disease in Hebei Province
Evidence regarding the short-term effects of ambient air pollution on acute exacerbations of chronic obstructive pulmonary disease (AECOPD) across multiple cities in northern China remains limited. This study aimed to evaluate the associations between major ambient air pollutants and AECOPD hospitalizations, as well as their lagged effects. We conducted an ecological time-series study of 14,786 AECOPD hospitalizations recorded at 49 hospitals in Shijiazhuang, Handan, and Chengde, Hebei Province, from January 1, 2018, to December 31, 2020. City-specific associations between daily concentrations of fine particulate matter (PM 2 . 5 ), inhalable particulate matter (PM 10 ), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), carbon monoxide (CO), and 8-hour ozone and AECOPD hospitalizations were estimated using distributed lag nonlinear models, with adjustment for long-term trends, temperature, relative humidity, day of the week, public holidays, and stages of the coronavirus disease 2019 pandemic. Single-day lag effects from lag 0 to lag 5, cumulative lag effects from lag 0–1 to lag 0–5, effects of extreme pollutant concentrations, and associations stratified by age and sex were evaluated. City-specific estimates were subsequently pooled using random-effects multivariate meta-analysis. Associations were expressed as relative risks with 95% confidence intervals. During the 1096-day study period, 14,786 AECOPD hospitalizations were recorded. Compared with the 25th-percentile concentrations, the 75th-percentile concentrations of PM 2.5 , PM 10 , SO 2 , NO 2 , and CO generally showed stronger cumulative associations with longer lag periods, with the largest estimates usually observed at lag 0–5 days, whereas single-day effects were weak. All 6 pollutants exhibited nonlinear cumulative exposure–response relationships. Compared with median concentrations, the 99th-percentile concentrations of PM 2.5 , PM 10 , SO 2 , NO 2 , and CO were associated with higher hospitalization risks. High PM 2.5 reached its maximum effect at lag 0–5 days (relative risk, 1.391; 95% confidence interval, 1.169–1.658). Higher 8-hour ozone concentrations were generally associated with lower hospitalization risk. Associations were more pronounced among adults aged ≥65 years and males. Short-term exposure to PM 2 . 5 , PM 10 , SO 2 , NO 2 , and CO was associated with increased AECOPD hospitalization risk, with nonlinear, lagged, and cumulative effects. Older adults and females may be more susceptible to ambient air pollution.
- 14 August 2026
- Published by Wolters Kluwer Health in Medicine
- Vol. 105 (33) , e50080
Open AccessAssociation between air pollution and acute exacerbation of chronic obstructive pulmonary disease in Hebei Province
Evidence regarding the short-term effects of ambient air pollution on acute exacerbations of chronic obstructive pulmonary disease (AECOPD) across multiple cities in northern China remains limited. This study aimed to evaluate the associations between major ambient air pollutants and AECOPD hospitalizations, as well as their lagged effects. We conducted an ecological time-series study of 14,786 AECOPD hospitalizations recorded at 49 hospitals in Shijiazhuang, Handan, and Chengde, Hebei Province, from January 1, 2018, to December 31, 2020. City-specific associations between daily concentrations of fine particulate matter (PM 2 . 5 ), inhalable particulate matter (PM 10 ), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), carbon monoxide (CO), and 8-hour ozone and AECOPD hospitalizations were estimated using distributed lag nonlinear models, with adjustment for long-term trends, temperature, relative humidity, day of the week, public holidays, and stages of the coronavirus disease 2019 pandemic. Single-day lag effects from lag 0 to lag 5, cumulative lag effects from lag 0–1 to lag 0–5, effects of extreme pollutant concentrations, and associations stratified by age and sex were evaluated. City-specific estimates were subsequently pooled using random-effects multivariate meta-analysis. Associations were expressed as relative risks with 95% confidence intervals. During the 1096-day study period, 14,786 AECOPD hospitalizations were recorded. Compared with the 25th-percentile concentrations, the 75th-percentile concentrations of PM 2.5 , PM 10 , SO 2 , NO 2 , and CO generally showed stronger cumulative associations with longer lag periods, with the largest estimates usually observed at lag 0–5 days, whereas single-day effects were weak. All 6 pollutants exhibited nonlinear cumulative exposure–response relationships. Compared with median concentrations, the 99th-percentile concentrations of PM 2.5 , PM 10 , SO 2 , NO 2 , and CO were associated with higher hospitalization risks. High PM 2.5 reached its maximum effect at lag 0–5 days (relative risk, 1.391; 95% confidence interval, 1.169–1.658). Higher 8-hour ozone concentrations were generally associated with lower hospitalization risk. Associations were more pronounced among adults aged ≥65 years and males. Short-term exposure to PM 2 . 5 , PM 10 , SO 2 , NO 2 , and CO was associated with increased AECOPD hospitalization risk, with nonlinear, lagged, and cumulative effects. Older adults and females may be more susceptible to ambient air pollution.
- 24 July 2026
- Published by Springer Nature in Applied Spatial Analysis and Policy
- Vol. 19 (3) , 177
Understanding the Association Between Neighbourhood Type, Built Environment, and Overweight in China: Evidence From Hebei Province
- 24 July 2026
- Published by Springer Nature in Applied Spatial Analysis and Policy
- Vol. 19 (3) , 177
Understanding the Association Between Neighbourhood Type, Built Environment, and Overweight in China: Evidence From Hebei Province
- 22 July 2026
- Published by Taylor & Francis in Blood Pressure
- p. 1-31
Association of air pollutant levels and blood pressure variability in patients with essential hypertension: a distributed lag analysis study
Environmental pollution and progressive climate change represent major contributors to the global health burden, causing over 9 million premature deaths annually. Among environmental exposures, the...
- 22 July 2026
- Published by Taylor & Francis in Blood Pressure
- p. 1-31
Association of air pollutant levels and blood pressure variability in patients with essential hypertension: a distributed lag analysis study
Environmental pollution and progressive climate change represent major contributors to the global health burden, causing over 9 million premature deaths annually. Among environmental exposures, the...
- Open Access
- 8 July 2026
- Published by Frontiers Media SA in Frontiers in Public Health
- Vol. 14, 1871195
Vitamin D deficiency among apparently healthy adults in northern China: behavioral correlates and an indoor-lifestyle framework
Background/objectives: Vitamin D deficiency is not confined to traditionally defined high-risk groups, yet its prevalence and modifiable behavioral correlates among apparently healthy adults remain poorly characterized, particularly in northern China. This study assessed vitamin D status among adults undergoing routine health examinations in Hebei Province and examined whether modifiable behavioral factors were independently associated with deficiency risk. Methods: A cross-sectional study was conducted among 4,811 adults attending routine health examinations in Hebei Province, China, after exclusion of participants with documented major chronic conditions known to affect vitamin D metabolism. Serum 25-hydroxyvitamin D [25(OH)D] was measured, and vitamin D deficiency was defined as <20 ng/mL. Structured questionnaires were used to collect information on sunlight exposure, physical activity, dietary vitamin D intake, and vitamin D-containing supplement use. Multivariable logistic regression models were applied to examine associations between these behavioral factors and vitamin D deficiency, with adjustment for sex, age group, smoking status, alcohol consumption, BMI category, and season of blood collection. A cumulative behavioral score was constructed to assess the combined association of favorable behavioral factors with deficiency risk. Results: The median serum 25(OH)D concentration was 20.5 ng/mL (IQR 15.4–26.6), and 47.5% of participants were vitamin D deficient, while only 15.3% had sufficient vitamin D status. Deficiency was more prevalent in women, younger and working-age adults, and participants examined during winter. In the fully adjusted model, sunlight exposure ≥30 min/day (OR 0.677, 95% CI 0.599–0.765), moderate-to-high physical activity (OR 0.773, 95% CI 0.678–0.880), adequate dietary vitamin D intake (OR 0.734, 95% CI 0.651–0.828), and vitamin D supplementation (OR 0.797, 95% CI 0.705–0.902) were each associated with lower odds of deficiency. A graded inverse association was observed for the cumulative behavioral score; compared with participants with no favorable behavioral factors, those with four had substantially lower odds of deficiency (OR 0.285, 95% CI 0.189–0.423). Conclusion: Vitamin D deficiency was highly prevalent among adults attending routine health examinations in northern China after exclusion of major chronic conditions known to affect vitamin D metabolism. The burden was concentrated in women, younger and working-age adults, and winter months, while lower deficiency risk was consistently associated with more favorable behavioral patterns. The observed distribution is consistent with an indoor-lifestyle framework in which vitamin D status is shaped not only by the environmental availability of sunlight, but also by the extent to which daily routines permit effective ultraviolet exposure.
- 8 July 2026
- Published by Frontiers Media SA in Frontiers in Public Health
- Vol. 14, 1871195
Open AccessVitamin D deficiency among apparently healthy adults in northern China: behavioral correlates and an indoor-lifestyle framework
Background/objectives: Vitamin D deficiency is not confined to traditionally defined high-risk groups, yet its prevalence and modifiable behavioral correlates among apparently healthy adults remain poorly characterized, particularly in northern China. This study assessed vitamin D status among adults undergoing routine health examinations in Hebei Province and examined whether modifiable behavioral factors were independently associated with deficiency risk. Methods: A cross-sectional study was conducted among 4,811 adults attending routine health examinations in Hebei Province, China, after exclusion of participants with documented major chronic conditions known to affect vitamin D metabolism. Serum 25-hydroxyvitamin D [25(OH)D] was measured, and vitamin D deficiency was defined as <20 ng/mL. Structured questionnaires were used to collect information on sunlight exposure, physical activity, dietary vitamin D intake, and vitamin D-containing supplement use. Multivariable logistic regression models were applied to examine associations between these behavioral factors and vitamin D deficiency, with adjustment for sex, age group, smoking status, alcohol consumption, BMI category, and season of blood collection. A cumulative behavioral score was constructed to assess the combined association of favorable behavioral factors with deficiency risk. Results: The median serum 25(OH)D concentration was 20.5 ng/mL (IQR 15.4–26.6), and 47.5% of participants were vitamin D deficient, while only 15.3% had sufficient vitamin D status. Deficiency was more prevalent in women, younger and working-age adults, and participants examined during winter. In the fully adjusted model, sunlight exposure ≥30 min/day (OR 0.677, 95% CI 0.599–0.765), moderate-to-high physical activity (OR 0.773, 95% CI 0.678–0.880), adequate dietary vitamin D intake (OR 0.734, 95% CI 0.651–0.828), and vitamin D supplementation (OR 0.797, 95% CI 0.705–0.902) were each associated with lower odds of deficiency. A graded inverse association was observed for the cumulative behavioral score; compared with participants with no favorable behavioral factors, those with four had substantially lower odds of deficiency (OR 0.285, 95% CI 0.189–0.423). Conclusion: Vitamin D deficiency was highly prevalent among adults attending routine health examinations in northern China after exclusion of major chronic conditions known to affect vitamin D metabolism. The burden was concentrated in women, younger and working-age adults, and winter months, while lower deficiency risk was consistently associated with more favorable behavioral patterns. The observed distribution is consistent with an indoor-lifestyle framework in which vitamin D status is shaped not only by the environmental availability of sunlight, but also by the extent to which daily routines permit effective ultraviolet exposure.
- Top Cited PapersOpen Access
- 1 June 2026
- Published by Taylor & Francis in Risk Management and Healthcare Policy
- Vol. ume 19, 1-11
Clinical Artificial Intelligence Implementation in Routine Care: Real-World Operational Outcomes from a Provincial Health System in China
Background: Clinical artificial intelligence (AI) technologies are increasingly being introduced into hospital practice, yet evidence describing their operational integration and performance after deployment in routine clinical settings remains limited. This study examined the real-world implementation and operational integration of clinical AI within a provincial tertiary health system in China over an 18-month observation period. Methods: This retrospective longitudinal observational study used aggregated institutional data generated during routine platform deployment, including electronic medical record–linked system logs, deployment records, quality-monitoring summaries, and operational reports. The analysis focused on implementation patterns, workflow integration, selected operational indicators, and user acceptance during routine clinical use. The study evaluated implementation and operational integration rather than algorithmic accuracy, diagnostic performance, or patient-level clinical effectiveness. Results: Three AI-supported clinical pathways were included: an intelligent pre-consultation system, a multidisciplinary tumor decision-support system, and a duloxetine therapeutic drug-monitoring pathway. During the observation period, 127 clinicians across 53 specialties participated in AI-assisted clinical activities involving more than 27,000 patient encounters, 850 multidisciplinary tumor decision-support cases, and 320 therapeutic drug-monitoring episodes. Patient waiting time decreased from 18 to 13 minutes, patient satisfaction increased from 95.40% to 98.92%, consultation efficiency improved by approximately 40%, and documentation completion efficiency improved by approximately 80%. Implementation patterns differed substantially across pathways, reflecting differences in workflow position and clinical accountability rather than deployment effort alone. Conclusion: In this provincial tertiary health system, clinical AI implementation was associated with sustained operational use, cross-specialty workflow integration, and measurable changes in selected workflow indicators. These findings suggest that technical functionality alone is insufficient for sustained clinical AI use and that workflow compatibility and organizational readiness are central to routine implementation. Because this was a single-site observational study using aggregated operational data, the findings should be interpreted as implementation evidence rather than proof of clinical effectiveness. Plain Language Summary: Many hospital AI systems are tested in pilot settings but are not subsequently integrated into routine clinical operations. This study examined how a clinical AI platform was deployed in routine care at a large teaching hospital in Hebei Province, China, over an 18-month period. The study used routine hospital data, including electronic medical record–linked system logs, deployment records, quality-monitoring summaries, operational reports, and patient satisfaction information. No identifiable patient information was accessed. The analysis examined how widely the AI tools were used, which clinical services adopted them, and whether selected workflow indicators changed following routine deployment. Three AI-supported tools were included: an intelligent pre-consultation system, a multidisciplinary tumor decision-support system, and a duloxetine therapeutic drug-monitoring system. In total, 127 clinicians across 53 specialties used the platform, involving more than 27,000 patient encounters, 850 tumor decision-support cases, and 320 drug-monitoring episodes. Patient waiting time decreased from 18 to 13 minutes, patient satisfaction increased from 95.40% to 98.92%, consultation efficiency improved by approximately 40%, and documentation efficiency improved by approximately 80%. The findings suggest that clinical AI can move from pilot use into routine hospital workflows when deployment is grounded in existing clinical workflows and backed by sustained organizational support. Because this was an observational study from a single hospital, the findings should be interpreted cautiously and confirmed in other settings.
- 1 June 2026
- Published by Taylor & Francis in Risk Management and Healthcare Policy
- Vol. ume 19, 1-11
Top Cited PapersOpen AccessClinical Artificial Intelligence Implementation in Routine Care: Real-World Operational Outcomes from a Provincial Health System in China
Background: Clinical artificial intelligence (AI) technologies are increasingly being introduced into hospital practice, yet evidence describing their operational integration and performance after deployment in routine clinical settings remains limited. This study examined the real-world implementation and operational integration of clinical AI within a provincial tertiary health system in China over an 18-month observation period. Methods: This retrospective longitudinal observational study used aggregated institutional data generated during routine platform deployment, including electronic medical record–linked system logs, deployment records, quality-monitoring summaries, and operational reports. The analysis focused on implementation patterns, workflow integration, selected operational indicators, and user acceptance during routine clinical use. The study evaluated implementation and operational integration rather than algorithmic accuracy, diagnostic performance, or patient-level clinical effectiveness. Results: Three AI-supported clinical pathways were included: an intelligent pre-consultation system, a multidisciplinary tumor decision-support system, and a duloxetine therapeutic drug-monitoring pathway. During the observation period, 127 clinicians across 53 specialties participated in AI-assisted clinical activities involving more than 27,000 patient encounters, 850 multidisciplinary tumor decision-support cases, and 320 therapeutic drug-monitoring episodes. Patient waiting time decreased from 18 to 13 minutes, patient satisfaction increased from 95.40% to 98.92%, consultation efficiency improved by approximately 40%, and documentation completion efficiency improved by approximately 80%. Implementation patterns differed substantially across pathways, reflecting differences in workflow position and clinical accountability rather than deployment effort alone. Conclusion: In this provincial tertiary health system, clinical AI implementation was associated with sustained operational use, cross-specialty workflow integration, and measurable changes in selected workflow indicators. These findings suggest that technical functionality alone is insufficient for sustained clinical AI use and that workflow compatibility and organizational readiness are central to routine implementation. Because this was a single-site observational study using aggregated operational data, the findings should be interpreted as implementation evidence rather than proof of clinical effectiveness. Plain Language Summary: Many hospital AI systems are tested in pilot settings but are not subsequently integrated into routine clinical operations. This study examined how a clinical AI platform was deployed in routine care at a large teaching hospital in Hebei Province, China, over an 18-month period. The study used routine hospital data, including electronic medical record–linked system logs, deployment records, quality-monitoring summaries, operational reports, and patient satisfaction information. No identifiable patient information was accessed. The analysis examined how widely the AI tools were used, which clinical services adopted them, and whether selected workflow indicators changed following routine deployment. Three AI-supported tools were included: an intelligent pre-consultation system, a multidisciplinary tumor decision-support system, and a duloxetine therapeutic drug-monitoring system. In total, 127 clinicians across 53 specialties used the platform, involving more than 27,000 patient encounters, 850 tumor decision-support cases, and 320 drug-monitoring episodes. Patient waiting time decreased from 18 to 13 minutes, patient satisfaction increased from 95.40% to 98.92%, consultation efficiency improved by approximately 40%, and documentation efficiency improved by approximately 80%. The findings suggest that clinical AI can move from pilot use into routine hospital workflows when deployment is grounded in existing clinical workflows and backed by sustained organizational support. Because this was an observational study from a single hospital, the findings should be interpreted cautiously and confirmed in other settings.
- Open Access
- 29 May 2026
- Vol. 16, 03014
When health systems invest matters: temporal imbalance and the case for life-course reorientation.
Health systems worldwide face growing pressure from population ageing and the rising burden of noncommunicable diseases. Yet policy debates still focus mainly on how much is spent, while paying much less attention to when health systems invest. In many settings, spending remains concentrated in late-stage curative care, whereas earlier phases of the life course, when risks are still developing and remain modifiable, receive less sustained investment. We describe this mismatch as a temporal imbalance: a misalignment between when health risks accumulate and when resources are allocated. We argue that the pattern persists because institutions are usually rewarded for responding to visible burden, while the returns to earlier intervention are delayed, diffuse, and harder to attribute. As a result, organisations that invest early often cannot readily claim the later benefits of doing so. This creates a temporal accountability gap. Drawing on life-course epidemiology, health financing research, clinical perspectives from endocrinology and palliative care, and illustrative expenditure data from Hebei Province, China, we examine timing as an overlooked dimension of health policy and propose three practical policy actions: routine life-course expenditure reporting, longer evaluation horizons for prevention, and financing mechanisms that reward earlier investment.
- 29 May 2026
- Vol. 16, 03014
Open AccessWhen health systems invest matters: temporal imbalance and the case for life-course reorientation.
Health systems worldwide face growing pressure from population ageing and the rising burden of noncommunicable diseases. Yet policy debates still focus mainly on how much is spent, while paying much less attention to when health systems invest. In many settings, spending remains concentrated in late-stage curative care, whereas earlier phases of the life course, when risks are still developing and remain modifiable, receive less sustained investment. We describe this mismatch as a temporal imbalance: a misalignment between when health risks accumulate and when resources are allocated. We argue that the pattern persists because institutions are usually rewarded for responding to visible burden, while the returns to earlier intervention are delayed, diffuse, and harder to attribute. As a result, organisations that invest early often cannot readily claim the later benefits of doing so. This creates a temporal accountability gap. Drawing on life-course epidemiology, health financing research, clinical perspectives from endocrinology and palliative care, and illustrative expenditure data from Hebei Province, China, we examine timing as an overlooked dimension of health policy and propose three practical policy actions: routine life-course expenditure reporting, longer evaluation horizons for prevention, and financing mechanisms that reward earlier investment.
- Top Cited PapersOpen Access
- 7 May 2026
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 28 (1) , e92680
From Pilot Trap to Institutional Capacity: A Governance Framework for Sustainable Clinical AI Implementation in Health Systems
Clinical artificial intelligence (AI) applications frequently fail to transition from short-term pilot projects into sustained components of routine clinical care, a phenomenon referred to in this viewpoint as the pilot trap. This persistent gap reflects not only technical or regulatory limitations but also insufficient governance capacity within health care organizations. We argue that such capacity is not fully established before deployment; rather, it develops through implementation as real-world operational tensions clarify organizational ownership, accountability boundaries, and coordination mechanisms. Drawing on an 18-month implementation of a provincial clinical AI platform in China, we develop a 6-module governance framework encompassing institutional carrier formation, infrastructure governance, regulatory and ethical governance, interdisciplinary coordination, translational scaling, and lifecycle evaluation and oversight. These modules represent functional governance conditions observed during implementation rather than a prescriptive institutional architecture to be installed prior to deployment. We further introduce the concept of functional transferability and position the framework as an upstream complement to existing international governance standards, which typically specify what governance should achieve, but often assume that the organizational capacity to implement it already exists. Advancing clinical AI beyond demonstration, therefore, depends less on model performance alone than on the ability of health systems to develop and sustain the institutional capacity required for routine clinical use.
- 7 May 2026
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 28 (1) , e92680
Top Cited PapersOpen AccessFrom Pilot Trap to Institutional Capacity: A Governance Framework for Sustainable Clinical AI Implementation in Health Systems
Clinical artificial intelligence (AI) applications frequently fail to transition from short-term pilot projects into sustained components of routine clinical care, a phenomenon referred to in this viewpoint as the pilot trap. This persistent gap reflects not only technical or regulatory limitations but also insufficient governance capacity within health care organizations. We argue that such capacity is not fully established before deployment; rather, it develops through implementation as real-world operational tensions clarify organizational ownership, accountability boundaries, and coordination mechanisms. Drawing on an 18-month implementation of a provincial clinical AI platform in China, we develop a 6-module governance framework encompassing institutional carrier formation, infrastructure governance, regulatory and ethical governance, interdisciplinary coordination, translational scaling, and lifecycle evaluation and oversight. These modules represent functional governance conditions observed during implementation rather than a prescriptive institutional architecture to be installed prior to deployment. We further introduce the concept of functional transferability and position the framework as an upstream complement to existing international governance standards, which typically specify what governance should achieve, but often assume that the organizational capacity to implement it already exists. Advancing clinical AI beyond demonstration, therefore, depends less on model performance alone than on the ability of health systems to develop and sustain the institutional capacity required for routine clinical use.
- 10 April 2026
- Published by Springer Nature in Journal of Perinatology
- Vol. 46 (10) , 1887-1893
Noninvasive oxygenation monitoring in PPHN: validation of the oxygen saturation index and the oxygen saturation to fraction of inspired oxygen ratio against conventional invasive indices
To evaluate the correlation and agreement between the Oxygen Saturation Index (OSI) and Oxygenation Index (OI), and between the SpO2/FiO2 ratio (S/F) and PaO2/FiO2 ratio (P/F) in neonates with persistent pulmonary hypertension (PPHN), and to identify noninvasive thresholds for clinical use. This retrospective study analyzed 2717 paired arterial blood gas and SpO2 measurements from 377 PPHN infants treated at a tertiary NICU (2012-2021). Correlation (Spearman), linear mixed-effects models (LMMs), Bland-Altman agreement, and ROC analyses were performed across SpO2 strata. OSI strongly correlated with OI (r = 0.93), and S/F with P/F (r = 0.84), with best performance at SpO2 ≥80%. Agreement declined under severe hypoxemia. ROC analysis showed excellent discrimination for OSI (AUC > 0.96) and good performance for S/F (AUC up to 0.96). OSI and S/F are reliable noninvasive surrogates for invasive oxygenation indices when SpO2 ≥ 80% and may help reduce arterial sampling in PPHN monitoring.
- 10 April 2026
- Published by Springer Nature in Journal of Perinatology
- Vol. 46 (10) , 1887-1893
Noninvasive oxygenation monitoring in PPHN: validation of the oxygen saturation index and the oxygen saturation to fraction of inspired oxygen ratio against conventional invasive indices
To evaluate the correlation and agreement between the Oxygen Saturation Index (OSI) and Oxygenation Index (OI), and between the SpO2/FiO2 ratio (S/F) and PaO2/FiO2 ratio (P/F) in neonates with persistent pulmonary hypertension (PPHN), and to identify noninvasive thresholds for clinical use. This retrospective study analyzed 2717 paired arterial blood gas and SpO2 measurements from 377 PPHN infants treated at a tertiary NICU (2012-2021). Correlation (Spearman), linear mixed-effects models (LMMs), Bland-Altman agreement, and ROC analyses were performed across SpO2 strata. OSI strongly correlated with OI (r = 0.93), and S/F with P/F (r = 0.84), with best performance at SpO2 ≥80%. Agreement declined under severe hypoxemia. ROC analysis showed excellent discrimination for OSI (AUC > 0.96) and good performance for S/F (AUC up to 0.96). OSI and S/F are reliable noninvasive surrogates for invasive oxygenation indices when SpO2 ≥ 80% and may help reduce arterial sampling in PPHN monitoring.