Ultrahuman M1 CGM Platform drives significant changes in Metabolic Score and eHbA1C in engaged participants of Ultra30 coaching program

Ria Pawar, Ved Asudani, Bhuvan Srinivasan, Mohit Kumar, Nishanth Krishnan, Aditi Bhattacharya
Background and rationale
Lifestyle management to achieve tangible physiological results has been prevalent for centuries. Dieting, exercise profiling and changes in sleep habits as guided management modules to achieve weight loss/ gain, stress reduction and enhance mobility to finally improve physical, and by extension mental well-being have experienced waves of evolution increasingly supported by scientific insights. The current wearable-driven, personalised lifestyle management has revolutionised goal-based lifestyle management by integrating  multi-faceted sensor data built on strong clinical study conclusions.  However, the majority of sensor data is derived from clinically-diagnosed populations, be that obesity, hypertension or type 2 diabetes mellitus (T2DM). Additionally, most short training bootcamps target a metabolically vulnerable group. Studies show promise, with a 2-day program improving waist circumference and body fat in individuals with Metabolic Syndrome1. However, data on bootcamps targeting healthy, and relatively younger individuals is scarce. This limits benchmarking of wearables catering to the healthy group and understanding of adherence/drop-out rates in a less physiologically constrained demographic.
Ultra30 bootcamps were conducted between September 2022 to January 2023 in three independent cohorts with self-declared healthy participants. Harnessing the power of the M1 continuous glucose monitoring (CGM) system and the clinically-validated Metabolic Score algorithm2, this free bootcamp series was offered to new Ultrahuman users or those already in possession CGM sensors for a 4-week period. The bootcamp primarily aimed at empowering users to make healthier choices by delivering structured tutorials on interpreting CGM data, decoding nutritional information found on food labels, and engaging in optimal exercise and fueling methodologies at a group-level (see timeline schematic below). The bootcamp content was created in-house and delivered by a trained fitness professional from the Ultrahuman team (UH-POC), who also served as the single point of contact for all participants. Though no personalised meal or fitness plans were created, participants could reach out to UH-POC for individual guidance or clarification, if required. There were no mandatory commitments and participants could discontinue at any time. While this was a bootcamp, the level of guidance from the UH-POC was relatively low compared to bootcamps that prescribe workouts and meals. It was largely self-directed with the main emphasis being on awareness and education.
General week on week flow of Bootcamp cohorts
Of a total of 800 respondents to the initial call for volunteers, 414 participants signed-up for the three bootcamps. These were self-declared healthy individuals who were not on any chronic medication or undergoing any supportive therapy that could influence the outcomes. Though the participants were requested to share age, BMI and other attributes, a negligible fraction complied. We found a 33.76% drop-out at the initial stage of CGM-activation, which was attributed to scheduling challenges given the fixed dates of bootcamp. Hence the number of active participants across the three cohorts were 308, with a final group of 133 participants who successfully completed the 4 week period, with a steep discontinuation noted at week 2 to week 3 transition.  Majority of the participants identified weight loss to be their prime driver to volunteer. A smaller subset reported improving productivity/work focus and sleep habits. A handful desired to increase their exercise stamina and performance.

Behavioural adherence and Metabolic awareness
The first two weeks focussed on enhancing the nutritional and exercise physiology awareness of the participants. Daily posts were communicated regarding the benefits of meal planning, sample menus, decoding food labels, activity fueling, understanding exercise types and timing nutrition with activity. Weeks 3-4 involved awareness of indirect physiological effects like sleep deficits, stress and emotional eating. The final week also incorporated a “challenge” which invited participants to apply the training of the preceding weeks to optimise their metabolic scores and communicate their outcomes on a shared , dedicated Slack®  channel. Organically collected user feedback from those who completed the bootcamps revealed that the majority of participants were unaware of correct nutritional design of their diets and paired with CGM, drove home the strong link of “eating right” and the power of small lifestyle adjustments to daily habits that can catalyse big changes in fitness.  Glucose utilisation post exercise also was found insightful and helped many participants to optimise their routines and allow for rest and recovery. Though the common Slack messaging channel was available for responses throughout the 4 weeks, participants were not that engaged and were hesitant to share their “hacks”. Though there were typically 8-10 posts from UH-POC in a day, these were not found to be an overload by most participants, indicating that app-based nudges are not disruptive when individuals are committed to a specific healthcare program.
Biomarker outcomes
Response across full cohort: An aggregate analysis of all participants who completed the bootcamp revealed a decrease in daily mean glucose values of 3.23 mg/dL across 4 weeks which was statistically significant (p<0.05, Repeated measures ANOVA, N=133). Metabolic score and time in range (70-110 mg/dL) did not vary significantly for the whole group and which is expected given the large glycemic heterogeneity observed across the group. Ultrahuman M1 platform also computes an estimated HbA1c (eHbA1c) based on weekly glucose dynamics. Comparing week 1 to week 4 paired data for participants revealed a downward trend of eHbA1c, but was not statistically significant (p=0.25).
Week on week average metrics - full cohort
Sustained results in the top 30th percentile cohort:  We compiled data of all participants who were in the top 30th percentile within the entire bootcamp period for each metric with a similar approach as mentioned above. These participants displayed high engagement as a sub-group with far fewer dropouts. This sub-group showed remarkable increases in metabolic score, reduction in average glucose and elevated time in ranges. The sub-group being relatively more homogenous, also yielded strong statistical significance between week 1 and week 4 in all metrics measured as shown in the graph below. Comparing week 1 to week 4 paired data for this group revealed a statistically significant decrease of 8% (0.45 points, p<0.001).
Week on week average metrics - Top 30th percentile
eHBA1c values for week 1 to week 4
Insights and takeaways
In the previously mentioned observational multivariate analysis3, we were able to find a clear difference in baseline, real-world levels of metabolic score and glycemic fitness in healthy and pre-diabetic Indians/SouthAsians. The question that remained was whether these metrics could be harnessed in an interventional mode for focussed fitness improvement regimes and whether tangible benefits could be derived in relatively short time intervals. Additionally, it was unclear how much real world engagement and adherence is found in a largely healthy demographic3,4. This latter feature is related to consistency and discipline that translates to
tangible results but is usually not used as a stratification tool in clinical or real-world data studies in the metabolic fitness domain5,6. We found that:
With the broadest inclusion criteria and design that allowed discontinuation at any time, we found a consistent decrease in average blood glucose across four weeks.
More than a third of the enrolled individuals did not engage, alluding the high barriers in daily life that inhibit commencing lifestyle changes. This supports the Ultrahuman approach of passive health monitoring and trying to minimise the effort required from the user. There is merit in examining adherence rates in larger cohorts and dissecting the factors driving drop-outs in such heterogeneous groups.
Among those who did complete three weeks or more of the bootcamps, it was a challenge to maintain and consistently prioritise healthy choices, which reflected in a flattened whole-group response over time.
Of those who did improve or showed higher engagement levels, there was far less variability of change in metabolic score and glucose values consistently, as indicated by the high statistical significance in the paired t-tests. This trend was also mirrored in the estimated HbA1c metric computed by the M1 platform based on the ADA formula7.
Since CGM provides a direct and dynamic insight into glucose levels, the effect is most pronounced in these metrics. User feedback did mention improved productivity and sleep patterns but the feedback received was for a small fraction of active participants.
In summary, this Ultra30 bootcamp did not focus on prescriptive food or exercise changes, but rather on improving awareness and user understanding of their own physiology and tools to measure it. Even in this mildly directive design, we found that for healthy Indians, significant improvements can be achieved in glucose metabolism (mean glucose, time in range, metabolic score, estimated HbA1c), provided there is a minimum of three to four weeks of engaged participation.

Future studies with improved sensors and data collection streams, coupled with multi-tiered analyses, is hoped to increase the amplitude of positive changes observed translating to  benefits in larger populations.
Ultrahuman offers bespoke training services via the Ultrahuman Rewire program. For more information go to www.ultrahuman.com or write to support@ultrahuman.com. For questions on this whitepaper contact: aditi@ultrahuman.com
© Ultrahuman 2024