Innovative monitoring in paediatrics in low-resource settings: an aid to save lives?

Innovative Monitoring in Paediatrics in Low-resource settings: In search of machine learning algorithms using vital signs and biomarkers

Status

Active, not recruiting

Phases

Unknown

Study type

Observational

Source

ISRCTN

Registry ID

ISRCTN71392921

Enrollment

1000

Registered

2022-04-25

Start date

2022-07-01

Completion date

Unknown

Last updated

2026-06-15

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Early detection and prevention of critical illness in children in high dependence unit (HDU) in Malawian hospitals Signs and Symptoms

Interventions

This is a clinical observational study aimed at the identification of predictors of critical illness in children using vital signs, clinical information, and point-of-care (POC) biomarkers at Queen El

Eligibility

Sex/Gender

All

Age

28 Days to 60 Months

Inclusion criteria

Inclusion criteria: Children aged 28 days – 60 months that will be admitted to the high dependency bays

Exclusion criteria

Exclusion criteria: 1. Children outside the age range 2. Children not admitted to the high dependency bays/ward 3. Children in which monitoring is technically not possible 4. No informed consent was given on admission

Design outcomes

Primary

Measure	Time frame
Critical Illness events are measured every 4 hours by the clinical hospital nurse: 1. Respiratory - Start or increase respiratory support: oxygen or CPAP, (Non)Invasive ventilation: bag & mask ventilation or intubation, start or increase of bronchodilator support 2. Circulatory - Transfusion of blood (products), Intravenous fluid bolus of 10ml/kg or more, start or increase of continuous/intermittent inotropic support (IV/IM adrenalin), cardio-pulmonary resuscitation (CPR): resuscitation setting involving chest compressions 3. Neurological - Decrease in Blantyre Coma Score of 1 point or more, Convulsions requiring anticonvulsants 4. Other 4.1. Sepsis: clinical suspicion of sepsis that has led to the collection of new blood culture and/or starts or change in antibiotic treatment 4.2. Start anti-malarial treatment 4.3. Objectified hypoglycaemia requiring correction (IV or enteral) 4.4. Unplanned admission to the (P)ICU 4.5. Unplanned surgical procedure (including chest drains) 4.6. Death	—

Measure

Time frame

Critical Illness events are measured every 4 hours by the clinical hospital nurse: 1. Respiratory - Start or increase respiratory support: oxygen or CPAP, (Non)Invasive ventilation: bag & mask ventilation or intubation, start or increase of bronchodilator support 2. Circulatory - Transfusion of blood (products), Intravenous fluid bolus of 10ml/kg or more, start or increase of continuous/intermittent inotropic support (IV/IM adrenalin), cardio-pulmonary resuscitation (CPR): resuscitation setting involving chest compressions 3. Neurological - Decrease in Blantyre Coma Score of 1 point or more, Convulsions requiring anticonvulsants 4. Other 4.1. Sepsis: clinical suspicion of sepsis that has led to the collection of new blood culture and/or starts or change in antibiotic treatment 4.2. Start anti-malarial treatment 4.3. Objectified hypoglycaemia requiring correction (IV or enteral) 4.4. Unplanned admission to the (P)ICU 4.5. Unplanned surgical procedure (including chest drains) 4.6. Death

—

Secondary

Measure	Time frame
1. To assess the contribution of adding sociodemographic data to the predictive power of the algorithms. 2. To assess the potential added role of a predefined set of biomarkers in a predictive algorithm to detect critical illness. These markers will be including POC testing of two RNA transcription markers (FAM89A and IFI44L), C-Reactive protein and further analysis including previously described markers. 3. Design separate predictive models for critical illness in a better-resourced centre (Blantyre) as compared to a less-resourced centre (Zomba) which may more closely resemble most sub-Saharan hospital settings and assess the accuracy of these models. 4. Design separate predictive models for critical illness in infants (aged between 28 days and 12 months) and older children (12 to 60 months) and assess the accuracy of these models. 5. Make predictive models using conventional statistical approaches to compare the accuracy (sensitivity, specificity, false-positive and negative ratios, and area under the curve) of our model(s) against models published in the literature or identified in a retrospective dataset.	—

Measure

Time frame

1. To assess the contribution of adding sociodemographic data to the predictive power of the algorithms. 2. To assess the potential added role of a predefined set of biomarkers in a predictive algorithm to detect critical illness. These markers will be including POC testing of two RNA transcription markers (FAM89A and IFI44L), C-Reactive protein and further analysis including previously described markers. 3. Design separate predictive models for critical illness in a better-resourced centre (Blantyre) as compared to a less-resourced centre (Zomba) which may more closely resemble most sub-Saharan hospital settings and assess the accuracy of these models. 4. Design separate predictive models for critical illness in infants (aged between 28 days and 12 months) and older children (12 to 60 months) and assess the accuracy of these models. 5. Make predictive models using conventional statistical approaches to compare the accuracy (sensitivity, specificity, false-positive and negative ratios, and area under the curve) of our model(s) against models published in the literature or identified in a retrospective dataset.

—

Countries

Malawi

Outcome results

None listed

Source: ISRCTN (via WHO ICTRP) · Data processed: Jun 21, 2026