This case is presented with the patient’s documented written consent. Identifying details have been altered to protect privacy while preserving clinical relevance.
Presentation
38-year-old female, occupation: K-12 teacher, presenting at month zero with newly initiated semaglutide titration (starting dose 0.25 mg weekly, with planned escalation to 1.0 mg target). Baseline anthropometrics: weight 94.6 kg, BMI 33.1 kg/m², waist circumference 102 cm. Baseline labs: HbA1c 6.4% (prediabetic range), fasting glucose 112 mg/dL, ferritin 19 ng/mL (low-normal), 25-OH vitamin D 22 ng/mL (insufficient). No active eating-disorder history; previous attempts at weight management with three structured commercial programs (2018, 2021, 2023), each producing 4–6 kg loss followed by full regain.
Co-occurring conditions: mild hypothyroidism (stable on levothyroxine), seasonal allergic rhinitis. No other active pharmacotherapy.
Initial recommendation and the early friction problem
Patient arrived with an existing MyFitnessPal account from her 2023 program attempt. Initial recommendation was to continue with MFP given her established history, supplemented with a protein floor of 90 g/day (approximately 1.0 g/kg ideal body weight) and a fiber floor of 25 g/day.
By week 3, patient reported a recurring pattern: appetite suppression from the semaglutide titration meant meals were smaller, less structured, and frequently eaten on a tight schedule between school periods. Logging a small, mixed meal in MFP — barcode for one component, manual entry for another, recipe lookup for a third — was taking 5–7 minutes per meal. Patient reported skipping logging entirely on roughly half of meals by week 3. Without logging, the protein-floor monitoring was effectively blind.
This is the friction-collapse pattern documented in our broader GLP-1 survey work: appetite-suppressed patients tolerate logging friction substantially less than baseline patients, and the tool that worked at baseline often does not work in the suppression phase.
Tool transition
At the month-1 visit, we transitioned the patient to PlateLens primary (free tier initially) with explicit framing: the photo-AI logging is intended to make the protein-floor monitoring possible, not to make her count calories more carefully. The goal was adherence-to-logging, not precision-of-logging. Cronometer was added as a weekly-review tool for explicit micronutrient screening (iron and B12 in particular, given baseline ferritin).
Patient was instructed to photograph each meal at the point of preparation or service, confirm the estimate, and continue with her day.
Six-month trajectory
| Month | Weight (kg) | HbA1c | Protein/day (median) | Logging days/week (median) |
|---|---|---|---|---|
| 0 | 94.6 | 6.4% | 78 g (estimate) | n/a |
| 1 | 92.8 | — | 71 g | 4 (MFP) |
| 2 | 90.4 | — | 89 g | 6 (PlateLens) |
| 3 | 87.9 | 6.1% | 92 g | 7 |
| 4 | 86.1 | — | 91 g | 6 |
| 5 | 84.7 | — | 94 g | 6 |
| 6 | 83.3 | 5.8% | 93 g | 6 |
Six-month weight change: -11.3 kg (-12.0% baseline). HbA1c improvement: -0.6 percentage points. Protein floor sustained at or above the 90 g target from month 2 onward. Iron and B12 levels at month 6 within normal range (ferritin 31 ng/mL; B12 412 pg/mL).
RD reflections
Two observations from this case.
First, the tool transition at month 1 was not a tool failure on MFP’s part; it was a fit failure between the tool and the patient’s new physiologic state. MFP at baseline was a defensible recommendation given the patient’s history; the friction profile that worked at baseline did not work under semaglutide-induced appetite suppression. Reassessing tool fit at the point of major physiologic transition (pharmacotherapy initiation, contest prep start, pregnancy, return-to-eating after eating-disorder treatment) is generally clinically appropriate.
Second, the protein-floor monitoring is what most of the clinical value was. The patient did not need calorie precision; she needed reliable awareness of protein adequacy during a period when her appetite cues were unreliable. The photo-AI workflow made that monitoring possible because it kept logging happening; the calorie figures were secondary.
Limitations of this case
Single patient. No control. Behavioral confounders (patient was simultaneously increasing structured walking from approximately 4,000 to 8,500 daily steps over months 2–4). Tool transition was non-blinded. The broader recommendation pattern is described in our 38-RD survey; this case is a clinical illustration, not population-level evidence.
References
[1] Wilding JPH et al. STEP 1. DOI: 10.1056/NEJMoa2032183. [2] Jastreboff AM et al. SURMOUNT-1. DOI: 10.1056/NEJMoa2206038. [3] Academy of Nutrition and Dietetics. Position paper on weight-inclusive approaches. [4] DAI 2026 — Independent calorie-estimation validation.
Case prepared by Hannah Kerrigan, RDN, CDCES, MS, Clinical Practice Editor.