Table 2

Summary of ML prediction studies reviewed

Author	Disease	Data source	Centres	Study dates	Age range	Primary outcome(s)	Missing data	ML method (best model)	Validation		Prospective study/RCT
Author	Disease	Data source	Centres	Study dates	Age range	Primary outcome(s)	Missing data	ML method (best model)	Internal	External	Prospective study/RCT
Hogan et al 202121	Asthma	Registry (National Database)	NR	2013	5–18	Hospitalisation	Complete case analysis	ANN	Yes	No	No
Filipow et al 202122	CF	Registry (Regional Database)	1	2000–2018	2–18	Hospitalisation, exacerbation	Complete case analysis	Cluster Analysis	Yes	Yes	No
Bose et al 202130	Asthma	Registry (EHR)	30+	2005–2016	2–5	Asthma persistence	Mixed	XGBoost	Yes	No	No
van Horck et al 202138	CF	Study data	3	NR	5–18	Exacerbation	NR	Random Forest	No	No	No
Raita et al 202125	Bronchi-olitis	Study data	17	2011–2014	<1	Wheeze, asthma	Imputed	Cluster Analysis	No	No	No
Sills et al 202126	Asthma	Registry (EHR)	5	2009–2013	2–21	Hospitalisation	Complete case analysis	AutoML	Yes	No	No
Seol et al 202129	Asthma	NR	1	NR	<18	Exacerbation	NR	Naïve Bayes	Yes	No	Yes
Lovrić et al 202142	Asthma	Study data	1	NR	2–22	Response to treatment	Imputed	AdaBoost	Yes	No	No
Caparrós-Martín et al 202044	CF	Study data	NR	NR	2.5–4.4	Lung damage (quantified CT)	NR	Cluster Analysis	No	No	No
Wang et al 201931	Asthma	Study data	8	1993–1995	5–12	Asthma remission	Complete case analysis	Decision Tree	No	No	No
Messinger et al 201932	Asthma	Registry (EHR)	1	2016–2017	2–18	Paediatric Asthma Score	Complete case analysis	ANN	Yes	No	No
Khasha et al 201945	Asthma	Study data	1	NR	>5	Asthma control	Imputed	Ensemble Learning	No	No	No
Goto et al 201933	Asthma	Registry (National Database)	NR	2007–2015	2–14 (IQR)	Critical care (admission to intensive care unit or death), hospitalisation	Imputed	Decision Tree	Yes	No	No
Patel et al 201834	Asthma	Registry (EHR)	2	2012–2015	2–18	Hospitalisation	Complete case analysis	Gradient Boosting	Yes	No	No
Ross et al 201835	Asthma	Study data	8	1993–1995	5–12	Asthma control	Imputed	Predictor Pursuit	Yes	No	No
Spyroglou et al 201840	Asthma	Study data	1	2008–2016	1–14.5	Exacerbation	NR	Bayesian	Yes	No	No
Huffaker et al 201827	Asthma	Study data	1	2016–2017	5–17	Exacerbation	Complete case analysis	Random Forest	Yes	No	No
Shin et al 201828	Asthma	Registry (EHR)	1	2016	6–18	Hospitalisation	NR	Random Forest	Yes	No	No
Das et al 201736	Asthma	Registry (EHR)	1	2013–2014	≤18	Frequent emergency department visits	NR	Logistic Regression	Yes	No	No
Pite et al 201639	Wheeze	Study data	1	1993–2006	≤7	Asthma development in adolescence	Complete case analysis	Cluster Analysis	No	No	No
Van Vliet et al 201643	Asthma	Study data	1	NR	6–18	Asthma control	NR	Random Forest	Yes	No	No
Luo et al 201537	Asthma	Study data	4	2011–2012	2–18	Asthma control 1 week prior	NR	Multiboost with Decision Stumps	Yes	No	No
Howrylak et al 201423	Asthma	Study data	8	1993–1995	5–12	Exacerbation	Imputed	Cluster Analysis	No	No	No
Farion et al 201324	Asthma	Registry (clinical records)	1	2000–2004	1–17	Exacerbation severity	Imputed	Naïve Bayes	Yes	No	Yes
Robroeks et al 201341	Asthma	Study data	1	NR	6–16	Exacerbation	NR	SVM	Yes	No	No

ANN, artificial neural network; EHR, electronic health record; NR, not reported; RCT, randomised controlled trial; SVM, support vector machine.