Sage Journals: Discover world-class research

Abstract

Introduction

Severe dengue and sepsis account for tens of millions of cases and deaths annually in low-resource settings, where spot-check vital sign monitoring misses early physiological deterioration. Affordable wearable sensors that stream continuous data and feed machine learning algorithms promise earlier detection and improved outcomes.

Methods

We conducted a structured narrative literature review of human studies on PubMed, Embase, Scopus, and Google Scholar that evaluated wearable vital-sign devices for predicting severe dengue or sepsis in resource-limited areas. Two reviewers independently screened, extracted, and synthesized data on study design, devices, vital streams, predictive performance, and implementation barriers.

Results

Seven eligible studies (2019–2024) from Vietnam, Rwanda, Bangladesh, and South Korea tested photoplethysmography or single-lead electrocardiography wearables in intensive care units and emergency departments of hospitals. The algorithms achieved an area under the receiver operating characteristic curve of 0.83–0.86 for forecasting dengue shock up to 2 hours and sepsis deterioration up to 5–48 hours before conventional recognition. The precision reached 0.79 for the three-level dengue severity classification. Continuous monitoring reduces the time to intervention and matches or exceeds the accuracy of manual charting.

Conclusion

Early evidence shows that low-cost wearable sensors can deliver clinically meaningful lead times for severe dengue and sepsis in developing countries’ hospitals. However, generalizability, artifact suppression, power autonomy, and economic sustainability remain unproven. Multicentre pragmatic trials, edge-optimized algorithms, and full cost-effectiveness analyses are needed before routine adoption.

Keywords

Wearable sensors vital-sign monitoring dengue sepsis low-resource settings early warning

Introduction

Dengue and sepsis, while distinct in their epidemiology and pathophysiology, impose overlapping and intensifying burdens on health systems worldwide. Dengue is a mosquito-borne viral disease endemic to tropical regions, often causing explosive outbreaks. In contrast, sepsis is a life-threatening syndrome triggered by dysregulated host responses to severe infections from various pathogens, typically presenting as sporadic cases rather than outbreaks.^1,2 Despite these differences, both conditions converge at the point of acute clinical deterioration, share overlapping physiological signatures (such as fever, hypotension, and shock), and disproportionately affect resource-limited low- and middle-income countries (LMICs). This shared clinical endpoint of rapid progression to critical illness provides a rationale for considering them together in evaluating wearable sensor-based early warning systems.

The World Health Organization (WHO) estimates that over 6.5 million cases of dengue and 7300 deaths were reported in 2023 alone, while in 2024, more than 14 million cases were reported with over 10,000 deaths, marking the highest number of global dengue cases ever recorded to date, thereby surpassing previous annual records.^3–5 In parallel, sepsis continues to exact a staggering toll, accounting for an estimated 48.9 million cases and 11 million deaths each year, representing roughly one-fifth of all global mortality, with nearly 20 million cases occurring in children under 5 years of age.⁶ Importantly, sepsis and dengue shock syndrome (DSS) exhibit several similar clinical features, which create a diagnostic dilemma, particularly in dengue-endemic regions where both conditions coexist.⁷ This underscores the need for rapid physiological monitoring to aid frontline clinicians in differentiating between early manifestations of dengue and sepsis. Evidence from an Indian case study highlights how wearable vital-sign sensors can support earlier recognition in these challenging scenarios.⁸

Recent advances in wearable vital-sign sensors offer a promising countermeasure: a low-cost wrist-mounted photoplethysmography (PPG) device that deployed predicted severe dengue deterioration 2 hours in advance with an area under the receiver operating characteristic curve (AUROC) of 0.83 and recall of 0.95, supporting real-time ward triage in a resource-constrained hospital.⁹ A gradient-boosted model using only six continuously streamed vitals from consumer-grade wearables forecasted sepsis onset up to 48 hours earlier with AUROCs of 0.84–0.83 across external test sets.¹⁰ Another study showed machine-learning–driven alerting systems reduced sepsis mortality by 44% (RR 0.56, 95% CI 0.39–0.80) compared with rule-based care.¹¹ Additional clinical deployment reports describe continuous ward monitoring via ViSi Mobile and similar platforms that shift observation from 4-hour spot-checks to minute-level streams, shortening antibiotic-to-door times and lowering sepsis deaths across 12 hospitals.¹²

Despite these advantages, critical gaps remain. Most sensor algorithms are validated in tertiary centers rather than primary care or community settings typical of dengue-endemic LMICs^9,13; head-to-head comparisons of signal quality loss during high humidity, extreme ambient heat, or pediatric shock are scarce^14,15; interoperability with minimal electronic infrastructures and cost-effectiveness over the device life cycle are understudied.^16,17 Furthermore, there is limited evidence on how real-time data streams integrate with frontline clinical workflows to trigger fluid resuscitation or vector-control responses in LMIC outbreaks.^18,19 Bridging these gaps is important because variables captured in our review tables such as heart-rate variability, pulse transit time, peripheral perfusion index, SpO₂, skin temperature, NEWS2/mSOFA scores, device cost, and battery autonomy, map directly onto outcomes that drive both scientific inquiry (algorithm generalizability), policy (procurement and reimbursement), and bedside practice (earlier intravenous fluid or platelet transfusion),²⁰ especially where clinician-to-patient ratios are low and laboratory turnaround is slow. Thus, this study aims to synthesize and critically appraise human studies of wearable vital-sign sensors for the early detection/prediction of severe dengue and sepsis in resource-limited settings, with a particular focus on predictive performance and implementation barriers in LMIC hospitals.

Methods

We conducted a structured narrative literature review to map the clinical evidence, challenges, and public health relevance of wearable vital sign sensors for the early detection of severe dengue and sepsis in low-resource settings. Although not a full systematic review, the methodology was strengthened by aligning with the elements of the PRISMA reporting framework to enhance transparency and reproducibility. PubMed, Embase, Scopus, and Google Scholar were searched for relevant studies. The search strategy used a combination of controlled vocabulary (e.g. Medical Subject Headings [MeSH]) and free-text keywords. Search strings combined exposure terms such as “wearable,” “biosensor,” “continuous monitoring,” AND and OR “photoplethysmography” with device-specific terms like “ECG patch,” paired with condition terms including “dengue,” “dengue shock,” “sepsis,” and “septic shock,” using Boolean operators AND and OR. Studies published after 2015 were prioritized. Eligible sources were peer-reviewed journal articles in English that enrolled humans and reported diagnostic, prognostic, or implementation outcomes of wearable vital sign sensors. Inclusion was restricted to studies conducted in low-resource or LMIC-relevant contexts, which were determined based on the study's geographic setting, explicit reference to resource-constrained environments, or descriptions of infrastructure limitations. We excluded editorials, commentaries, conference abstracts without full data, and studies that did not focus on dengue, sepsis, or LMIC-relevant contexts from the review. Additional exclusion criteria included studies limited to simulation or bench testing without clinical deployment and those evaluating non-wearable devices (e.g. handheld monitors or fixed bedside equipment). Two reviewers (MMA and ZKO) independently screened the titles, abstracts, and full texts, with disagreements resolved through discussion with the third reviewer (OJO). Data from the included studies were extracted into standardized evidence tables capturing the characteristics of the included studies, devices, clinical settings, monitored vital signs, target conditions, predictive performance, usability, and key limitations of wearable approaches. The findings were narratively synthesized into domains covering study characteristics across the included investigations, wearable sensor platforms, care settings, physiological streams, model precision, clinical lead time, and implementation challenges.

Results

Study characteristics across included investigations

Seven eligible studies were published between 2019 and 2024 and represented Vietnam, Rwanda, Bangladesh, and South Korea, with Vietnam contributing half of the total.^9,13,18,^21–24 Prospective observational designs predominated (six of seven, 85%), while one study used a prospective cohort framework, underscoring a descriptive emphasis in early stage validation.^9,13,18,^21–24 Dengue-focused work accounted for three Vietnamese studies (Table 1), whereas sepsis motivated the remaining five.^9,18

Table 1.

Characteristics of included studies.

Author and year	Reference	Country	Study objective(s)	Study design
Karolcik et al. (2024)	²¹	Vietnam	Assess a low-cost wrist photoplethysmography (PPG) wearable for automated dengue-severity stratification	Prospective observational
Ming et al. (2024)	⁹	Vietnam	Forecast impending dengue shock in real time from continuous PPG signals	Prospective observational
Trieu et al. (2022)	²²	Vietnam	Use compensatory reserve index derived from pulse-oximetry waveform to predict recurrent shock in dengue	Prospective observational
Garbern et al. (2019)	¹³	Rwanda	Validate a chest-patch biosensor for continuous vitals in emergency sepsis patients	Prospective observational
Garbern et al. (2024)	¹⁸	Bangladesh	Test wearable and machine learning alert system for very early detection of advanced pediatric sepsis	Prospective observational
Ghiasi et al. (2022)	²³	Vietnam	Compare 24-hours single-lead electrocardiography wearable with bedside monitors for sepsis-mortality prediction	Prospective observational
Choi et al. (2022)	²⁴	South Korea	Evaluate wireless heart rate/respiratory rate monitoring for earlier septic-shock prediction in febrile emergency patients	Prospective cohort

Wearable sensor platforms, care settings, and physiological streams

All deployments occurred in hospital environments, including five intensive care units and two high-acuity emergency departments, reflecting a preference for closely monitored clinical contexts during early implementation^13,18 (Table 2). Photoplethysmography wrist or fingertip devices were used exclusively in dengue investigations, whereas chest or forehead electrocardiography (ECG) patches formed the backbone of sepsis monitoring; a pattern more likely reflecting the specific expertise and established practices of Vietnamese dengue studies rather than intrinsic disease-specific physiology.^21,23 Heart rate was captured in every dataset, respiratory rate in five studies, peripheral oxygen saturation or compensatory reserve indices in three studies, and skin temperature in two studies, producing multimodal vital-sign streams of varying breadth.²² Across diseases, each team pursued an early warning paradigm: dengue studies aimed to stratify severity or forecast shock up to 2 hours in advance, whereas sepsis investigations sought to detect advanced disease, short-term mortality several hours before conventional assessments.^9,24

Table 2.

Devices, clinical settings, monitored vital signs, and target conditions.

Author and year	Reference	Setting/data source	Wearable/sensor	Vital signs	Target condition
Karolcik et al. (2024)	²¹	Hospital for tropical diseases	SmartCare Analytics wrist wearable finger PPG sensor	Pulse-rate waveform Continuous PPG waveforms	Dengue severity class
Ming et al. (2024)	⁹	Hospital for tropical Diseases	SmartCare Analytics wrist wearable finger PPG probe	Continuous PPG signals	Dengue shock (2 hours horizon)
Trieu et al. (2022)	²²	Pediatric and adult intensive care unit (ICU), tertiary referral hospital	Finger pulse-oximetry (CRI algorithm)	CRI, HR, and blood pressure (BP) trend	Recurrent dengue shock
Garbern et al. (2019)	¹³	University teaching-hospital emergency department	VitalPatch chest patch	Continuous HR, RR, and skin temperature	Sepsis
Garbern et al. (2024)	¹⁸	Pediatric ICU, Dhaka tertiary children's hospital	VitalPatch + smartphone	HR, RR, skin temperature, and SpO₂	Advanced pediatric sepsis
Ghiasi et al. (2022)	²³	Adult ICU, hospital for tropical diseases	ePatch single-lead ECG	HR and HR-variability	24-hour sepsis mortality
Choi et al. (2022)	²⁴	Isolation-care unit, tertiary emergency department	HiCardi ECG patch	HR and RR	Septic shock in febrile emergency patients

CRI: compensatory reserve index; PPG: photoplethysmography; HR: heart rate; RR: respiratory rate; ECG: electrocardiography.

Model accuracy, clinical lead time, and implementation challenges

The discrimination metrics were uniformly strong. A convolutional neural network applied to SmartCare wristband signals classified three dengue severity levels with a precision of 0.79²¹ (Table 3). A multimodal transformer predicted hemodynamic collapse 2 hours ahead of time with an AUROC of 0.83.⁹ A compensatory reserve index threshold below 0.40 signaled recurrent dengue shock within 12 hours, achieving a sensitivity of 0.66 and specificity of 0.86.²² Sepsis models showed comparable accuracy: ridge regression analysis of VitalPatch data in a pediatric intensive care unit reached an AUROC of 0.86 and delivered alerts a median of 2 hours earlier than clinical recognition.¹⁸ Gradient-boosted ECG features detected septic shock 5–9 hours sooner than intermittent observations and slightly outperformed manual charting (AUROC 0.86 vs. 0.84).²⁴ Agreement studies corroborated sensor validity, with heart rate correlation of r = 0.87 and respiratory rate correlation of r = 0.74 against manual reference in a Rwandan emergency cohort.¹³ Recurring operational barriers included small sample sizes (Figure 1), up to 40% signal loss from motion artifacts, battery or connectivity constraints, and requirements for specialized algorithms or staff training, indicating common implementation hurdles across low-resource hospitals.²³

Figure 1.

Operational barriers are reported across wearable-sensor studies.

Table 3.

Predictive performance, usability, and key limitations of wearable approaches.

Author and year	Reference	Analytical approach	Key performance metric	Reported clinical value	Principal limitations
Karolcik et al. (2024)	²¹	Convolutional neural network	Precision ≈ 0.79 for three-level classification	Automates ward triage and personalized monitoring	Small cohort; motion artefact; no real-world validation
Ming et al. (2024)	⁹	Multi-modal transformer	AUROC 0.83 for stable-state prediction	Enables safe outpatient management and bed-capacity relief	High data volume and battery demand; signal noise
Trieu et al. (2022)	²²	CRI < 0.4 threshold	Sens 0.66/Spec 0.86 for re-shock ≤ 12 hours	Alerts ICU staff to volume depletion for timely fluids	∼40% data loss; specialized algorithm
Garbern et al. (2019)	¹³	Agreement study (no machine learning)	HR r = 0.87; RR r = 0.74 versus manual	Feasible continuous vitals in resource-limited settings	Detachment; connectivity; staff training
Garbern et al. (2024)	¹⁸	Ridge-regression machine learning	AUROC 0.86; wearable alerts ≈ 2 h earlier than clinical recognition	Earlier escalation could improve outcomes in LMIC PICUs	Requires stable power/Internet; single centre; small cohort
Ghiasi et al. (2022)	²³	Recurrent neural network	AUPRC 0.83 (wearable data)	Automated early warning of high-risk sepsis patients	24-hours recording limit; small sample; specialized patch
Choi et al. (2022)	²⁴	Gradient boosting	AUROC 0.86 (wearable) versus 0.84 (manual)	Detected deterioration 5–9 hours earlier than intermittent vitals	False-alarm burden; device removal for imaging

CRI: compensatory reserve index; AUROC: area under the receiver operating characteristic curve; HR: heart rate; RR: respiratory rate; ICU: intensive care unit; LMICs: low- and middle-income countries; PICUs: psychiatric intensive care unit.

Discussion

Our synthesis shows that across diverse low-resource hospitals, the reviewed wearables consistently provided clinically useful lead times and strong discrimination for incipient shock or advanced sepsis, enabling earlier escalation of care than conventional four-hourly spot-checks. The physiology behind this performance is plausible: dengue plasma leakage and early sepsis vasodilation first alter pulse-pressure amplitude, heart-rate variability, peripheral perfusion index, and skin-temperature gradients, signals that photoplethysmography and single-lead ECG patches capture at sub-minute resolution, while transformer and gradient-boosting algorithms exploit the resulting high-frequency dynamics to detect decompensation well before absolute thresholds are breached.^25,26 In addition, one feasibility pilot study demonstrated the practicality of coupling a medical-grade wrist sensor to an artificial intelligence–enhanced NEWS2 engine for bedside monitoring, which identified sepsis deterioration hours earlier than standard nursing rounds.²⁷

These findings align with a propensity-matched analysis of 7971 medical ward admissions, where continuous wireless monitoring cut the combined odds of ICU transfer or in-hospital death by nearly threefold, underscoring how uninterrupted vital-stream analytics translate into real-world outcome benefits.²⁸ Moreover, cost and sustainability concerns are mitigated by emerging sub-US$150 prototypes whose PPG-based blood pressure trend and SpO₂ estimates met clinical accuracy thresholds in volunteer trials, suggesting that device affordability no longer precludes adoption in lower-income settings.²⁹ However, the literature remains skewed toward tertiary ICUs; a study found that the validation of wearable data flows into frontline decision-making and interoperability with sparse electronic infrastructures are critical gaps for broad implementation.³⁰

The main limitations of the studies center on study design, signal fidelity, power management, and economic uncertainty. Most were single-centered investigations with small cohorts (fewer than 200 participants). Half of the included studies originated from a single country (Vietnam), which raises the possibility of selection bias and constrains external validity. Continuous PPG and ECG streams are highly vulnerable to motion artifacts.^31,32 Laboratory tests that used multichannel wavelength filtering still lost accuracy during ambulation, and clinical studies reported discarding up to 43% of raw epochs because of poor signal quality. Power constraints further limited continuity.³³ The low-cost wrist system that predicted dengue shock required recharging after a median of 18 hours, creating unavoidable surveillance gaps. Algorithmic generalizability is also uncertain because models are usually trained and tested on data from the same site,³⁴ and only one large propensity-matched ward study has linked continuous monitoring to lower intensive care transfers and mortality, leaving causality provisional. Finally, none of the trials presented full cost-effectiveness analyses that accounted for device amortization, connectivity fees, consumables, or staff training; therefore, sustainability in low-resource budgets remains unproven.

To move the field forward, multicenter pragmatic trials that include district hospitals and community fever clinics should externally validate algorithms, document signal dropout rates, battery endurance and lifecycle costs, and report clinical and economic end points (Figure 2).³⁵ Developing edge-optimized models that run on on-device microcontrollers, together with robust motion-suppression pipelines and energy-harvesting or ultra-low-power chipsets, would extend operating time and reduce dependence on cloud bandwidth.³⁶ Parallel health-economic evaluations should incorporate procurement, maintenance, and training costs so that ministries of health can negotiate pooled tenders or local assembly schemes.³⁷ Interoperability and data-governance standards ought to align with the WHO Regional Digital Health Action Plan 2023–2030, which explicitly calls for evidence-based scaling of remote patient monitoring and secure stewardship of health data.³⁸ At the service level, hospitals can begin by embedding validated wearable dashboards into existing early warning systems, training nurses in sensor placement and escalation protocols, and establishing local maintenance pathways. Policymakers should update national sepsis and dengue guidelines to include wearable-enabled monitoring and create financing mechanisms that subsidize devices for frontline facilities. Together, these actions help convert continuous wearable surveillance from a promising research prototype into a reliable, sustainable safety net across LMICs’ health systems.

Figure 2.

Strategies for wearable surveillance integration.

Conclusion

Across the included studies, continuous photoplethysmography and ECG wearables consistently detected impending dengue shock or sepsis deterioration hours ahead of standard four-hourly observations, demonstrating AUROCs above 0.80 and translating to faster escalation of care. These gains plausibly reflect the devices’ ability to capture rapid fluctuations in pulse amplitude, heart rate variability, perfusion index, and skin temperature gradients that preceded over hemodynamic collapse. Nonetheless, evidence is still largely confined to small, single-center cohorts in tertiary hospitals; high motion-related signal loss, limited battery life, and uncertain lifecycle costs threaten scalability; and few investigations evaluate integration with frontline workflows or health-economic impact. To move from promising prototypes to sustainable clinical tools, future research must undertake multicenter trials in district hospitals and community fever clinics, rigorously benchmark artifact-suppression pipelines and on-device inference against resource constraints, and embed cost-effectiveness analyses that include procurement, maintenance, and training.

Footnotes

Acknowledgements

The authors acknowledge the use of Paperpal AI () for its “Language Edit” and “Make Academic” features to improve clarity and readability. This assistance was limited to linguistic refinement; all analyses and interpretations are solely those of the authors.

ORCID iDs

Mohamed Mustaf Ahmed

Zhinya Kawa Othman

Sumaya Mahamud Mahamed

Olalekan John Okesanya

Shuaibu Saidu Musa

Francesco Branda

Don Eliseo Lucero-Prisno III

Ethical approval

Not required for this study.

Consent to participate

Not applicable.

Consent to publication

Not applicable.

Author contributions

MMA conceptualized and designed the study. ZKO and SMM performed the literature review and data curation. MMA and OJO drafted the manuscript. SSM and FB revised the manuscript. DELP III supervised the study. All authors have read and approved the final manuscript.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

Not applicable; no new datasets were generated or analyzed.

Clinical trial registration

Not applicable.

References

Bola Olajide

Olajide

Olasehinde

. General perspectives on dengue fever. Clin Microbiol Newsl 2024; 49: 1–9.

Cao

Wang

Xie

. Immune dysregulation in sepsis: experiences, lessons and perspectives. Cell Death Discov 2023; 9: 465.

World Health Organization. Dengue - global situation, https://www.who.int/emergencies/disease-outbreak-news/item/2024-DON518 (2024, accessed 29 September 2025).

WHO. Dengue, https://www.who.int/news-room/fact-sheets/detail/dengue-and-severe-dengue (2025, accessed 29 September 2025).

Haider

Hasan

Onyango

, et al. Global dengue epidemic worsens with record 14 million cases and 9000 deaths reported in 2024. Int J Infect Dis 2025; 158: 107940.

WHO. Sepsis, https://www.who.int/news-room/fact-sheets/detail/sepsis (2024, accessed 6 July 2025).

Singh

Tiwari

Satone

, et al. Updates in the management of dengue shock syndrome: a comprehensive review. Cureus 2023. DOI: 10.7759/cureus.46713.

Yadav

Kamath

Gangane

, et al. Unmasking the mimic: when sepsis disguises as dengue shock syndrome. Int J Reprod Contracept Obstet Gynecol 2024; 13: 765–767.

Ming

Daniels

Chanh

, et al. Predicting deterioration in dengue using a low cost wearable for continuous clinical monitoring. NPJ Digit Med 2024; 7: 306.

10.

Barton

Chettipally

Zhou

, et al. Evaluation of a machine learning algorithm for up to 48-hour advance prediction of sepsis using six vital signs. Comput Biol Med 2019; 109: 79–84.

11.

Zhang

Chen

, et al. Effectiveness of automated alerting system compared to usual care for the management of sepsis. NPJ Digit Med 2022; 5: 101.

12.

Goldberg

Kao

Kwan

, et al. UCHealth’s virtual health center: how Colorado’s largest health system creates and integrates technology into patient care. NPJ Digit Med 2024; 7: 187.

13.

Garbern

Mbanjumucyo

Umuhoza

, et al. Validation of a wearable biosensor device for vital sign monitoring in septic emergency department patients in Rwanda. Digit Health 2019; 5. DOI: 10.1177/2055207619879349.

14.

Fine

Branan

Rodriguez

, et al. Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring. Biosensors (Basel) 2021; 11: 126.

15.

Fleischhauer

Bruhn

Rasche

, et al. Photoplethysmography upon cold stress—impact of measurement site and acquisition mode. Front Physiol 2023; 14. DOI: 10.3389/fphys.2023.1127624.

16.

Azodo

Williams

Sheikh

, et al. Opportunities and challenges surrounding the use of data from wearable sensor devices in health care: qualitative interview study. J Med Internet Res 2020; 22: e19542.

17.

Canali

Schiaffonati

Aliverti

. Challenges and recommendations for wearable devices in digital health: data quality, interoperability, health equity, fairness. PLOS Digit Health 2022; 1: e0000104.

18.

Garbern

Mamun

GMDS

Shaima

, et al. A novel digital health approach to improving global pediatric sepsis care in Bangladesh using wearable technology and machine learning. PLOS Digit Health 2024; 3: e0000634.

19.

Gathright

Mejia

Gonzalez

, et al. Overview of wearable healthcare devices for clinical decision support in the prehospital setting. Sensors 2024; 24: 8204.

20.

Armañac-Julián

Kontaxis

Rapalis

, et al. Reliability of pulse photoplethysmography sensors: coverage using different setups and body locations. Front Electron 2022; 3. DOI: 10.3389/felec.2022.906324.

21.

Karolcik

Manginas

Chanh

, et al. Towards a machine-learning assisted non-invasive classification of dengue severity using wearable PPG data: a prospective clinical study. EBioMedicine 2024; 104: 105164.

22.

Trieu

Khanh

Ming

DKY

, et al. The compensatory reserve index predicts recurrent shock in patients with severe dengue. BMC Med 2022; 20: 109.

23.

Ghiasi

Zhu

, et al. Sepsis mortality prediction using wearable monitoring in low–middle income countries. Sensors 2022; 22: 3866.

24.

Choi

Chung

, et al. Advantage of vital sign monitoring using a wireless wearable device for predicting septic shock in febrile patients in the emergency department: a machine learning-based analysis. Sensors 2022; 22: 7054.

25.

Lam

McBride

DHT

, et al. Visual and biochemical evidence of glycocalyx disruption in human dengue infection, and association with plasma leakage severity. Front Med (Lausanne) 2020; 7. DOI: 10.3389/fmed.2020.545813.

26.

Aguilar-Briseño

Moser

Rodenhuis-Zybert

. Understanding immunopathology of severe dengue: lessons learnt from sepsis. Curr Opin Virol 2020; 43: 41–49.

27.

Kuhn

S-O

Gibb

Gründling

. The role of the Corsano CardioWatch in continuous vital sign monitoring for early sepsis detection. Crit Care 2025; 29: 144.

28.

Rowland

Saha

Motamedi

, et al. Impact on patient outcomes of continuous vital sign monitoring on medical wards: propensity-matched analysis. J Med Internet Res 2025; 27: e66347.

29.

Filgueiras

Bertemes-Filho

Noveletto

. Evaluating the accuracy of low-cost wearable sensors for healthcare monitoring. Micromachines (Basel) 2025; 16: 791.

30.

Angelucci

Greco

Cecconi

, et al. Wearable devices for patient monitoring in the intensive care unit. Intensive Care Med Exp 2025; 13: 26.

31.

Zhang

Song

Vullings

, et al. Motion artifact reduction for Wrist-Worn photoplethysmograph sensors based on different wavelengths. Sensors 2019; 19: 673.

32.

Lee

Kim

Park

H-K

, et al. Motion artifact reduction in wearable photoplethysmography based on multi-channel sensors with multiple wavelengths. Sensors 2020; 20: 1493.

33.

Ebrahimi

Gosselin

. Ultralow-power photoplethysmography (PPG) sensors: a methodological review. |IEEE Sens J 2023; 23: 16467–16480.

34.

Fernstad

Svennberg

Åberg

, et al. External validation of a machine learning-based classification algorithm for ambulatory heart rhythm diagnostics in pericardioversion atrial fibrillation patients using smartphone photoplethysmography: the SMARTBEATS-ALGO study. Europace 2025; 27. DOI: 10.1093/europace/euaf031.

35.

Weinfurt

Hernandez

Coronado

, et al. Pragmatic clinical trials embedded in healthcare systems: generalizable lessons from the NIH Collaboratory. BMC Med Res Methodol 2017; 17: 144.

36.

Hizem

Bousbia

Ben Dhiab

, et al. Reliable ECG anomaly detection on edge devices for Internet of Medical Things applications. Sensors 2025; 25: 2496.

37.

Zhang

Weaver

Armstrong

, et al. Validity of wrist-worn photoplethysmography devices to measure heart rate: a systematic review and meta-analysis. J Sports Sci 2020; 38: 2021–2034.

38.

WHO. The ongoing journey to commitment and transformation, https://cdn.who.int/media/docs/librariesprovider2/data-and-evidence/english-ddh-260823_7amcet.pdf?download=true&sfvrsn=4c674522_2 (2023, accessed 6 July 2025).

Wearable vital-sign sensors for early detection of severe dengue and sepsis: Clinical applications and implementation in resource-limited settings

Abstract

Introduction

Methods

Results

Conclusion

Keywords

Introduction

Methods

Results

Study characteristics across included investigations

Wearable sensor platforms, care settings, and physiological streams

Model accuracy, clinical lead time, and implementation challenges

Discussion

Conclusion

Footnotes

Acknowledgements

ORCID iDs

Ethical approval

Consent to participate

Consent to publication

Author contributions

Funding

Declaration of conflicting interests

Data availability statement

Clinical trial registration

References