Table 1.

Baseline characteristics, grouped according to the groups used for the deep learning model.

	Patients training	Controls training	Patients validation	Controls validation	Sig.
Participants, n (% male)	50 (38%)	50 (38%)	20 (35%)	19 (42%)	–
Age, mean ± SD	54.7 ± 18.6 b	49.4 ± 15.1 b	55.3 ± 14.6 c	48.1 ± 11.8 c	0.910 a ; 0.731 a
Disease duration in years, median [IQR]	5.8 [2.4–18.8]	–	4.5 [1.8–19.2]	–	0.687
Antibodies, %		–		–	–
AChR	72%		90%
MuSK	14%		5%
Seronegative	14%		5%
Total QMG score, mean ± SD	9.0 ± 4.8	–	11.0 ± 6.7	–	0.274
Missing, n	1		2
Facial QMG score, mean ± SD	2.8 ± 2.4		3.4 ± 3.4		0.493
Facial QMG score 0–1 points, %	38%		41%
Mild MG (QMG 0–9)
N, %	29 (59%)		10 (56%)
QMG, mean ± SD	5.6 ± 2.3		6.0 ± 1.9		0.648
MGFA A/B, %	3%/34%		20%/10%
Moderate–severe MG (QMG >9)
N, %	20 (41%)		8 (44%)
QMG, mean ± SD	13.8 ± 3.2		17.5 ± 4.2		0.019
MGFA A/B, %	20%/65%		25%/63%
Cumulative prednisone dosage (mg) past 6 months, median [IQR]		–		–
Mild	827.5 [0–2151] d		0 [0–0] e		0.003
Moderate–severe	1471.6 [0–3758] d		1362.5 [0–3356] e		0.833

Bold values represent p‐values <0.05.

MG, myasthenia gravis; MGFA, myasthenia gravis foundation of America (A = limb predominant, B = bulbar predominant) at the time of the video recording; QMG, quantitative myasthenia gravis score.

^a First value represents between patient sets, second value is between control sets.

^b No significant difference in age between patients vs controls in training set: p = 0.117.

^c No significant difference in age between patients vs controls in validation set: p = 0.099.

^d No significant difference in prednisone dosage between mild and moderate–severe patients in training set: p = 0.323.

^e No significant difference in prednisone dosage between mild and moderate–severe patients in the validation set: p = 0.079.