Taking my diabetes treatment into my own hands

Original Article ↗ Hacker News Discussion ↗

Managing type 1 diabetes increasingly falls to patients themselves, who juggle insulin, diet, exercise and imperfect medical guidance with the help of continuous glucose monitors and, in some cases, DIY “artificial pancreas” systems. Commenters share how they model glucose–insulin dynamics, use open-source tools, closed‑loop pumps and data logging to tame the “all vibes” nature of dosing, while also describing serious risks like nocturnal hypoglycemia and diabetic ketoacidosis. The conversation widens to critique gaps in mainstream diabetes care, debate low‑carb and ketogenic diets (especially for type 2), and highlight how technology and self‑education can dramatically improve quality of life—but never fully remove the danger.

Adult-Onset T1D and Autoimmunity

  • Multiple accounts of “late” Type 1 onset, sometimes coinciding with other autoimmune diseases (e.g., rheumatoid arthritis).
  • Commenters note adult-onset T1D is common and was a driver for renaming “juvenile diabetes” to Type 1.
  • Viral infections (Epstein–Barr, CMV, flu) are mentioned as suspected triggers, but mechanisms are acknowledged as complex and not fully understood.

DIY Modeling and Optimization

  • The blog’s use of biophysical glucose–insulin models and open‑source libraries sparked debate.
  • Some argue using differential-equation models as black boxes without understanding them is risky; others point out this still exceeds typical clinical practice.
  • Suggestions for better optimization: treat doses as continuous variables; use derivative‑free / black‑box optimization (e.g., Bayesian optimization, standard numerical methods) instead of brute‑force genetic algorithms.
  • Probabilistic programming tools (PyMC, Stan) are mentioned for parameter estimation and uncertainty, but seen as an advanced topic.

Everyday Management Strategies

  • Strong support for pre‑bolusing ~15 minutes before eating; several T1Ds report dramatically smoother post‑meal glucose, despite clinicians sometimes downplaying it due to practical risks.
  • Additional tactics: walking after meals, splitting basal doses, extending boluses for fat/protein, confirming CGM extremes with fingersticks.
  • Emotional burden is a recurring theme: constant decision‑making, “vibes‑based” dosing, and periodic “screw it” moments around food.

Closed-Loop / Artificial Pancreas Systems

  • Several commenters already use commercial closed-loop systems (Medtronic, Tandem, Omnipod + Dexcom) and DIY setups (Loop, AndroidAPS, iAPS), often reporting life‑changing improvements in time‑in‑range and mental health.
  • Distribution is uneven: easier access in some US/UK settings than elsewhere; regulatory and reimbursement barriers remain.
  • Limitations noted: alarm fatigue, CGM inaccuracies, limited algorithm flexibility, UX issues; some still prefer DIY loops for configurability and sensor overlap.

Diet, Exercise, and T2D / Prediabetes

  • Many T2D and prediabetic commenters report major benefits or remission from low‑carb or ketogenic diets, sometimes combined with metformin or GLP‑1 drugs; others succeed on whole‑food, high‑carb plant‑based diets.
  • Broad agreement that weight loss, intense and regular exercise, and reducing fast carbs improve insulin sensitivity.
  • Multiple people stress that T1D absolutely still requires insulin, even on strict keto; attempts to replace insulin with diet alone are described as dangerous.
  • Some mention specific adjuncts (oats/beta‑glucan, turmeric, psyllium), but evidence quality is mixed and often anecdotal.

Risks: Hypoglycemia, DKA, and CGM Limits

  • Stories of nocturnal hypoglycemic coma and diabetic ketoacidosis highlight life‑threatening risks, especially during strenuous trips (heat/cold damaging insulin, lack of carbs, limited monitoring).
  • Concerns about CGM accuracy (false lows/highs), especially with “no calibration” sensors, fuel skepticism about fully automated dosing.
  • Alarm fatigue leads some to disable alerts, consciously shifting responsibility back to manual checks.

Healthcare System and Self-Advocacy

  • Strong consensus that T1Ds must learn to manage themselves; many feel routine care is too infrequent, formulaic, or tech‑illiterate to handle real‑world variability.
  • Some defend clinicians as overworked and under‑resourced; others describe shallow diagnostics, dismissal, and reliance on outdated care pathways.
  • Disappointment with “sick‑care” models drives people toward self‑logging, CGMs, DIY tools, and extensive personal research.