The Dexcom G7 and t:slim closed loop work well enough on a normal day. I’ve had type 1 diabetes for nearly a decade now, and the system is good — good enough that most days it fades into the background.
A competition day does not behave like a normal day. I see this from two overlapping contexts: autocross driving, and hillclimb event work. Both compress stress, heat, timing, and unpredictability into a system the closed loop was not explicitly designed to model.
What the loop assumes
Closed-loop insulin systems are designed around predictable relationships between glucose, insulin, food, and activity. They handle small variance well: missed snacks, delayed meals, or mild activity changes.
They do not model short, repeated adrenaline spikes. An autocross run is brief but intense — a sharp physiological load that begins before the car even moves and ends after it is already over.
Adrenaline independently raises glucose. The system responds after the fact, not in anticipation.
Driving vs working events
Autocross introduces repeated short stress cycles: run, cool down, reset, repeat. The glucose profile reflects spikes tied to discrete, predictable intervals.
Hillclimb event work is structurally different. It produces sustained low-level stress: long periods on foot, irregular breaks, variable meal timing, and reduced opportunities for self-monitoring.
The difference is not intensity alone, but predictability and control over timing.
The heat constraint
Heat affects both insulin behavior and hardware reliability.
In-car conditions introduce short, high-intensity heat exposure under full protective gear. That accelerates adhesive failure for both CGM sensors and infusion sites.
Outdoor event work produces longer-duration heat exposure with fewer natural reset points, shifting the problem from acute failure to gradual degradation.
Extra adhesive reinforcement is treated as baseline equipment, not contingency. It is applied before exposure, not after failure.
Timing and constraint pressure
On-track timing is externally controlled. Run order does not adjust for physiological state. Decisions must therefore be made in advance of data confirmation.
This shifts management from reactive correction to predictive adjustment: treating CGM trends as inputs, not alerts.
On event workdays, constraints are structural rather than scheduled, requiring deliberate check-in points rather than relying on natural breaks.
What actually works
Several consistent operational patterns emerged over time:
Front-loaded planning: insulin, meals, and adjustments are prepared relative to the expected structure of the day rather than adjusted ad hoc.
Event-aware correction strategy: post-run or post-stress corrections are treated as probabilistic rather than precise, reducing overcorrection cycles.
Intentional monitoring windows: checks are embedded into workflow gaps rather than waiting for alerts.
Redundant adhesion strategy: sensor and infusion security is treated as required equipment for all conditions.
Physiological awareness over device reliance: CGM data is treated as lagging indicator, not ground truth in isolation.
System limits
There is no configuration that fully resolves the interaction between adrenaline, heat, and external timing constraints in motorsports environments.
The system improves through management strategy, not elimination of variability.
The practical shift is from reacting to events to anticipating structural load before it occurs.
Conclusion
The most significant change over time has not been technological. It has been operational.
Closed-loop systems are effective within their design assumptions. Motorsports conditions sit partially outside those assumptions.
Managing that gap is less about optimization and more about discipline: planning ahead, reducing surprise, and treating the body as a system operating under known constraints rather than unpredictable failure.