The investigation evaluated if symptom signatures of myasthenia gravis exacerbations could be evaluated using real-world data gathered via a smartphone-based research platform.
The experiences of patients living with myasthenia gravis (MG) could soon benefit from digital phenotyping of their symptoms, with real-world, patient-reported data gathered from a smartphone-based research platform also helping to refine and predict potential MG exacerbations.
Of the original 113-patient population recruited from 37 states, 82 completed the 3-month study.1 Their mean (SD) age was 53.6 (14.0) years, 60% were female patients, and 87% reported a White ethnicity. Additional ethnicities reported were Black or African American (5%), Hispanic or Latinx (3%), and Asian, American Indian or Alaska Native, and Other (2% each). All participants had to submit proof of an MG diagnosis, and they were classified into 5 groups based on medication regimen: no reported active medication (group 0), symptomatic therapy (group 1), pyridostigmine and glucocorticoids (group 2), steroid-sparing chronic immunosuppression (group 3), and treatment for refractory disease (group 4). They also were tasked with completing daily check-ins in which they reported and rated MG symptoms using the MG-Activities of Daily Living (MG-ADL) scale (0 [normal] to 3 [severe]) and indicated if they thought an MG exacerbation was imminent.
“It can be challenging to study the presence of symptom exacerbations and MG crisis, due to the low prevalence of the condition,” investigators wrote in Frontiers in Neurology.2 “The ubiquity of smartphones and wearable devices allows data to be collected more frequently and passively as compared to traditional, site-based clinical studies, creating a more complete picture of the lived experience of the disease.”
The most common MG phenotype reported was severe MG among those who completed the study, and of them, 84% said they had several exacerbations each year. When they reported a low symptom burden, the median baseline MG-ADL score was 5 and for high symptom burdens, 14. Their most common comorbidity was hypertension (31.7%).
From the more than 4000 data points collected, 98% of the participants had days without exacerbations, 55% reported days with exacerbations, and 73% said they were unsure if they had any exacerbations. There were an average 6.3 exacerbations per participant for the entire study. Further, median (IQR) MG-ADL scores were higher during reported exacerbation periods vs nonexacerbation periods, at 7 (4-9) vs 0.3 (0.0-0.8), and the investigators saw a significant association between MG-ADL scores abd exacerbation status (Wilcoxon signed-rank value of P = 1.25 x 10-8).
Among the patients who reported their step counts (n = 26), the study authors saw that fewer steps were taken on days with exacerbations; there was significant variation in the average daily step count and exacerbation status, the authors noted. However, there was a weak correlation between daily step count and MG-ADL score (Pearson correlation coefficient r = −0.14), they added.
Among the symptoms reported by the study population, exacerbation days saw higher rates of difficulty swallowing (odds ratio [OR], 1.28) and impaired speech (OR, 1.34), and nonexacerbation days had higher rates of drooping eyelids (OR, 1.25) and leg (O/E ratio, 1.34) and arm weakness (OR, 1.22). An additional generalized linear mixed model incorporated data on symptom combinations and self-reported exacerbation status from 29 patients who reported exacerbations and were highly adherent to study tasks, and it found difficulty swallowing (OR, 2.67), impaired speech (OR, 2.44), shortness of breath (OR, 1.68), and blurred/double vision (OR, 1.39) were more likely to occur during an exacerbation. Also with this model, drooping eyelids (OR, 0.75) and leg weakness OR, 0.67) were more likely to occur absent an exacerbation.
When the data were examined according to medication group, group 4 (treatment for refractory disease) had a strong positive correlation with reporting exacerbations and groups 0 (no reported active medication) and 3 (steroid-sparing chronic immunosuppression) had strong negative correlations with exacerbation occurrence.
“The results summarized above suggest that some participant-reported data streams may be useful as features for development of a composite model to predict oncoming exacerbations in MG patients,” the authors concluded. “They suggest that digital phenotyping, characterized by increased multidimensionality and frequency of the data collection, holds promise for furthering our understanding of clinically significant exacerbations and reimagining the approach to treating MG as a ‘snowflake’ condition.”
Limitations on these findings include that they may not be generalizable to a broader population who has MG or to those with lower levels of digital literacy. Also, the authors noted the lack of an ethnically diverse study population.
1. Help Build an A.I. Model to Predict Myasthenia Gravis Symptom Patterns and Flares. NCT04590716. Updated July 29, 2021. Accessed September 6, 2023. https://clinicaltrials.gov/study/NCT04590716
2. Steyaert S, Lootus M, Sarabu C, et al. A decentralized, prospective, observational study to collect real-world data from patients with myasthenia gravis using smartphones. Front Neurol. 2023;14:1144183. doi:10.3389/fneur.2023.1144183