Machine learning approaches classify clinical malaria outcomes based on haematological parameters

Table 1 Characteristics of study participants for nMI, UM, and SM (n = 2,207)

Characteristic	Non-malaria infections (nMI)		Uncomplicated malaria (UM)		Severe malaria (SM)
N = 2,207	N = 978	(44.3%)	N = 703	(31.8%)	N = 526	(23.8%)	P value
Patient age
Mean (SD)	4.23	(3.57)	4.95	(3.57)	1.66	(0.93)	< 0.001^a
Median (range)	3.0	(2–6)	4.0	(2–7)	1.0	(1–2)
Body temperature
Mean (SD)	37.4	(1.18)	38.1	(1.23)	38.4	(1.15)	< 0.001^a
Median (range)	37.2	(36.5–38.4)	38.1	(37–39)	38.3	(37.5–39.2)
Parasite density
Geometric mean (SD)	0	0	27,467.59	8.44	16,674.41	8.61	0.592^c
Median (range)	0	0	29,426	3,144–105,351	25,160	3,560–86,560
Location
Accra (n, %)	657	(67.2%)	200	(28.4%)	0	(0.0%)	< 0.001^b
Kintampo (n, %)	321	(32.8%)	405	(57.6%)	0	(0.0%)
Navrongo (n, %)	0.0	(0.0%)	98	(13.9%)	526	(100.0%)
Sex
Female (n, %)	477	(48.8%)	317	(45.1%)	236	(44.9%)	0.209 ^b
Male (n, %)	501	(51.2%)	386	(54.9%)	290	(55.1%)
Fever symptom
No (n, %)	395	(40.4%)	97	(13.8%)	4	(0.8%)	< 0.001 ^b
Yes (n, %)	581	(59.4%)	601	(85.5%)	522	(99.2%)
Missing (n, %)	2	(0.2%)	5	(0.7%)	0	(0%)

Patient age, body temperature, and parasite density were analysed using the Kruskal-Wallis test while recruitment location, sex, and fever were analysed using the chi-square test at 95% CI. All the participant characteristics were significantly different between the nMI, UM, and SM except median parasite density and patient sex
^aKruskal-Wallis test
^bChi-square test
^cDunn (1964) Kruskal-Wallis multiple comparison—UM vs SM only

ISSN: 1741-7015