Explore how Smart QI Prediction Models are transforming healthcare quality improvement in low-resource settings, and learn about innovative research led by the Institute for Global Health at BC Children’s Hospital and BC Women’s Hospital + Health Centre to enhance patient outcomes through predictive analytics in global health.

Smart QI Prediction Models

Researchers at the Institute for Global Health at BC Children’s and Women’s Hospital, the University of British Columbia, WALIMU, Kenya Medical Research Institute, Rwanda Pediatric Association, iStreams and many other partners have worked collaboratively over the past decade to develop and validate data-driven risk prediction models to improve childhood mortality and morbidity outcomes in triage and discharge in low resource settings.

This suite of tools, called Smart Quality Improvement (QI), includes:

  • Smart Triage (0-12 years)
  • Smart Discharges
    • 0-6 months
    • 6 months - 5 years 
    • 5-16 years (under development)
    • 5-13 years within a Refugee cohort
    • Mother and newborn dyads

Prediction models have diagnostic and prognostic capabilities to inform decision making. They are developed to help health care providers in estimating the probability or risk of a specific disease or condition being present or of a specific event occurring in the future. 

Smart QI has developed digital adaptation kits (DAKs) following the WHO SMART guidelines, to facilitate integration and adoption of Smart Triage and Smart Discharges prediction models with existing digital systems. Paper-based systems are also being developed.  

Digital Adaptation Kit for Smart Triage: model and the independent danger signs (Web Annex C) 

Digital Adaptation Kit for Smart Discharge: models and risk score calculation can be found in Web Annex E. 

Smart Triage (0-12 years)

The Smart Triage predictive model is incorporated into a mobile app that allows healthcare providers to triage pediatric patients efficiently and effectively. 

The model uses nine predictor variables to place patients into three triage categories: emergency, priority, and non-urgent. In addition, independent danger signs are embedded in the digital health app that will immediately escalate a patient to a more urgent category. Predictors were selected through a rigorous process with several model development stages. Evaluation is ongoing to ensure feasibility and reproducibility.  

Relevant publications

Pre-development:

Development:

Feasibility Study:

Pivotal Study Protocol:

Validation:

Implication:

Smart Discharges

0-5 years

The Smart Discharges prediction models are based on candidate predictors collected at admission and used to predict mortality six months post-discharge. 

Three models were derived for each age group 0-6-month-olds and 6-60-month-olds. Each model was restricted to eight variables drawing from a different pool of available predictors. The different models utilize: 

  1. Commonly-available clinical variables; 
  2. Commonly-available clinical and social variables; and
  3. Any of the candidate predictor variables. 

Relevant publications:

Pre-development:

Development:

Validation:

Mother and newborn dyads

Pre-development:

5-16 years

Under development

5-13 years within a Refugee cohort

Under development

Our Team

Dr. Samuel Akech – Principal Investigator, KEMRI-Wellcome Trust Research Programme

Dr. Mark Ansermino – Executive Medical Director, Global Health; Co-Director, Digital Health Innovation Lab at BC Children’s; Professor, UBC Department of Anesthesia, Pharmacology and Therapeutics

Dr. Stephen Businge – Holy Innocents Children’s Hospital, Uganda

Dr. Guy Dumont – Investigator, BC Children’s Hospital; Co-Director, Digital Health Innovation Lab at BC Children’s; Professor, UBC Department of Electrical and Computer Engineering

Dr. Tumubugane Gotharido – Director, Uganda Martyrs Hospital, Ibanda 

Dr. Anneka Hooft - Assistant Clinical Professor, University of California San Francisco (UCSF); Pediatric Emergency Care Physician, UCSF Benioff Children’s Hospitals

Dr. Jerome Kabakyenga – Director, Maternal, Newborn and Child Health Institute at Mbarara University of Science and Technology  

Dr. Ronald Kasyaba – Assistant Executive Secretary, Uganda Catholic Medical Bureau

Dr. David Kimutai – Paediatrician, Mbagathi County Hospital Kenya

Dr. Niranjan Kissoon – Professor, UBC Departments of Pediatrics and Surgery (Emergency Medicine)

Dr. Aaron E Kornblith – Physician, UCSF Benioff Children’s Hospitals

Dr. Elias Kumbakumba – Physician/Researcher, Mbarara Regional Referral Hospital; Associate Professor, Mbarara University of Science and Technology Faculty of Medicine

Dr. Alfred Lumala – Director, St Josephs Hospital, Kitovu 

Dr. Nathan Kenya Mugisha – Executive Director, WALIMU

Dr. Angela Namala, Senior Gynaecologist, Jinja Regional Referral Hospital

Dr. Joseph Ngonzi – Dean, Faculty of Medicine, Mbarara University of Science and Technology, Senior Gynaecologist, Mbarara Regional Referral Hospital

Dr. Charles Olaro, Director Curative Services, Uganda Ministry of Health

Dr. Benard Opar – Director of Programs, WALIMU, Uganda

Dr. Sam Orach – Executive Secretary, Uganda Catholic Medical Bureau

Dr. Florence Oyella – Head of Department Paediatrics, Gulu Regional Referral Hospital

Dr. Jesca Nsungwa Sabiiti, Commissioner for Maternal and Child Health, Ugandan Ministry of Health 

Dr. Barak St John – Smart QI Clinical Lead, Kisiizi Hospital

Dr. Emmanuel Tenywa – Head of Department, Paediatrics, Jinja Regional Referral Hospital 

Dr. Abner Tagoola – Co-PI, Paediatrics, Jinja Regional Referral Hospital

Dr. Christian Umuhoza – Senior Lecturer College of Medicine and Health Sciences, University of Rwanda; 

Head of Pediatrics Emergency Unit, University Teaching Hospital of Kigali (CHUK); 

Dr. Emmanuel Uwiragiye – Physician, Ruhengeri Referral Hospital

Dr. Matthew O. Wiens – Investigator, IGH; Assistant Professor, UBC Department of Anesthesia, Pharmacology and Therapeutics

Partners

Smart QI Prediction Models partners

Funders

Smart QI Prediction Models funders