Dear Coaches,
This week we examine research testing whether AI can actually assess athlete nutrition, share the latest industry opportunities, and introduce a new solid food option for high-carbohydrate demands.
🧠EXPERT OPINION
AI Nutrition
Athletes are already asking ChatGPT what to eat.
The question is whether they should trust the answer.
Researchers from the University of Gothenburg tested three AI models on a straightforward task.
Take a photo of a meal, estimate the calories and macros.
They analysed 52 standardised food photographs with known weights and nutritional content.
ChatGPT and Claude achieved roughly 36% error rates for weight and energy estimation.
That's comparable to traditional self-reported dietary methods, which show 20-50% error rates in validation studies.
Minimal recall bias, reporting fatigue, and participant burden.
But here's the problem.
All three models systematically underestimated larger portions. The bigger the meal, the worse the estimate. For athletes with high energy demands, this creates a significant issue.
Macronutrient estimation was messier. Protein estimates exceeded 60% error. When Gemini misidentified falafel as meatballs, it overestimated protein by 360%.
Claude confused scrambled eggs for pasta, overestimating carbohydrates by 1,788%.
The study design itself revealed limitations. Vegetables were consistently placed in front, increasingly obscuring the calorie-dense components (starches and proteins) as portion sizes grew.
The models had progressively less visual information exactly where it mattered most.
Current general-purpose AI models show promise as screening tools but remain unsuitable for precise assessment in athletic populations requiring accurate quantification.
For general awareness or pattern tracking in recreational populations, the technology is approaching viability.
📈NEWS
Liam Howells joins Sunderland FC.
My Vital Metrics is raising on Crowdcude.
Tottenham opens Academy Nutritionist position.
HPSNZ seeks Performance Nutritionist, fixed term.
pn-insider+ drop new AI prompts for nutrition coaching.
Hurricanes rugby hiring Lead Performance Nutritionist.
Hexis seeks performance solutions consultant (€120k OTE).
Science in Sport hiring multiple sport nutrition content roles.
💡PERFORMANCE TECH
SiS Energy Oat Bar
Real ingredients meet high-carbohydrate demands.
Your athletes face a daily challenge: consuming 6-8g/kg carbohydrate when training schedules leave little time for proper meals.
Each 70g bar delivers 40g carbohydrates from oats.
Consume between meals to bridge carbohydrate gaps, pre-exercise for carbohydrate loading, or post-exercise when full meals aren’t immediately available.
Why this matters for your athletes:
Helps meet aggressive daily carbohydrate targets without relying on gels or powders.
Portable format eliminates meal timing stress during back-to-back training sessions or travel days.
Solid food option for athletes who struggle with gel-only fuelling strategies beyond 90 minutes.
