I am building a thing

So I've been feeding my blood results, eating, workouts into CoPilot... but it starts suffering from short term memory loss pretty fast.
I figured I can rather save things like workout details, blood tests, personal details (weight, height etc) into a db and use that as "context" to ask the LLM to evaluate stuff.

Log workouts, can do "Analyse my progress" across one or multiple workouts.

There's a consolidated view showing all journal entries (which could be anything from eating patterns to ... medication ... taken), bloodwork, workouts:

Once again... you can run analysis across any or all of the entries, giving a checkpoint.

The output is pretty cool... it's currently suggesting a deload.. lol. (The bloodwork isn't pulling in correctly yet... so apparently I'm "dead" according to the analysis... loooool).

 Health analysis by GoogleGemini (gemini-3.1-flash-lite-preview) on 03 Apr 2026 21:52
Disclaimer: I am an AI, not a doctor or healthcare professional. The insights provided here are for educational and informational purposes only and do not constitute medical advice, diagnosis, or treatment. Please consult with your physician or a qualified healthcare provider regarding your blood test results and any changes to your health or exercise regimen.

---
 1. Overall Health Snapshot
You are demonstrating a high level of dedication to physical conditioning, with a consistent and progressive training load. Your nutritional intake is structured, and you are tracking metrics with impressive detail. However, there are significant clinical red flags in your recent blood work that require immediate medical attention to ensure your internal physiology matches your external athletic effort.2. Blood Work Insights
The results from March 6th, 2026, contain several values that are physically impossible or indicate a major reporting/sampling error:
* S-BICARBONATE (0.0 mmol/L): A value of 0.0 is incompatible with life. This suggests either a laboratory error, a collection error, or an issue with the reporting system.
* S-CREATININE (110 umol/L): This is slightly above the reference range (64–104). While common in individuals with high muscle mass and intense resistance training, it should be discussed with a doctor to ensure it is due to muscle turnover and not kidney strain.
* Missing Data: The absence of Cholesterol, Triglycerides, and Urea markers makes it difficult to assess your metabolic or cardiovascular health.

Recommendation: Contact the PathCare laboratory to verify these results. If these are indeed accurate readings, you should see a physician immediately. If they are errors, request a re-test.3. Fitness Assessment
Your training consistency is excellent. You are following a logical progression of "progressive overload," as evidenced by the increase in weight on your Leg Press (80kg to 190kg) and consistent use of compound lifts.
* Progression: You are effectively increasing intensity while maintaining high volume.
* Exercise Selection: Your routine is well-rounded, focusing on major muscle groups.
* Caution: Your "Calories Burned" estimates from weight sessions are quite high; be aware that these trackers often overestimate caloric expenditure. Use these numbers as a trend, not an absolute truth.4. Lifestyle & Wellbeing
Your caloric and macronutrient tracking (late March/early April) shows a disciplined approach.
* Caloric Intake: You are fluctuating between ~1,600 and 2,100 calories. Depending on your height, weight, and body composition goals, you may be in a significant caloric deficit.
* Protein: You are averaging roughly 1.5–2g of protein per kg of body weight, which is excellent for muscle retention and repair.
* Wellbeing: You did not record subjective feelings or sleep data. To complete your holistic picture, try adding brief notes on "Energy levels," "Sleep quality (hours)," and "Perceived Stress" to your daily logs.5. Correlations & Connections
* Performance vs. Nutrition: On days where your caloric intake drops toward 1,500–1,600kcal (e.g., March 26th, April 1st), monitor if you experience increased fatigue during your workouts.
* Creatinine & Hydration: Increased physical intensity can raise creatinine levels. Ensure you are adequately hydrated, as dehydration can artificially elevate creatinine and complicate blood work interpretation.
* The "Gap" in Data: Your blood work (March 6th) and your nutrition tracking (March 24th–April 2nd) occur during different periods. It is possible that your nutritional status during the blood test was different from your current tracking phase.6. Actionable Recommendations 

• Medical Follow-up: Bring your March 6th report to your GP. Address the "0.0 mmol/L" Bicarbonate reading immediately, as this is the highest priority.
• Verify Creatinine: Discuss your high-volume weightlifting routine with your doctor to determine if your elevated creatinine is benign (related to exercise) or requires further kidney function investigation (like a Cystatin C test).
• Refine Subjective Tracking: Add a 1–10 scale for "Sleep Quality" and "Recovery/Soreness" to your journal entries. This will help you correlate your training volume with how your body actually feels.
• Hydration: Maintain consistent water intake, especially on the days you perform high-intensity training, to support renal health and metabolic function.
• Standardize Testing: In the future, try to schedule your blood tests under similar conditions (e.g., always in the morning, 24–48 hours after your last intense workout) to ensure the data is consistent and comparable.

Medical Follow-up: Bring your March 6th report to your GP. Address the "0.0 mmol/L" Bicarbonate reading immediately, as this is the highest priority. <- LOOOOL... how... I'm mos dead. lol

But anyway... I figured this might be interesting for you guys to see.

I follow numerous doctors in the blood work field and I work with 2 for my blood work business. AI is getting it very wrong for many blood markers.

Also their treatment plans are way off even using "hypotheticals" as many times pharmaceuticals need to be prescribed and Ai isnt even aware of the pharmaceutical options to fix issues and then correlating the potential side effects to blood work and contra indications of meds and nutracuticals..

Hence why no one else has done this task prior.

Ive been looking at clients blood work since 2014 as its been a major part of my coaching, and Ai wont correlate all the issues and I will give you a client case.

Client came to me with serveer anxiety and insomnia, been to numerous specialists and put close to 107k into doctors scopes etc. AI Spat out a treatment plan that made things even worse, even suggesting certain meds which ended up with him becoming anemic and needing 2 iron infusions 6 weeks apart.

The 2 doctors and I ended up figuring out that the use of certain medications for bacteria overgrowth in his gut( which was confirmed with scopes) altered his gut microbiome and stomach acidity.

He put the suggested treatment in to AI and it told him the treatment was "hearsay" and not evidence based..

Well, he followed the prescribed advice and took the meds prescribed we are now on day 67, of 6-7 hours of sleep aswell as day 31 off all anxiety meds with 90% improvement, and has iron levels coming back up to normal.

AI doesnt do well of figuring out optimal levels.

It's always interesting to hear from someone "in the field" doing stuff.

Personally I'm using what the LLMs spit out but only in conjunction with what my doctor advises... like I'm not about to inject drain cleaner because the LLM picked up an opinion piece that some crackpot put out... this is definitely a VERY real risk with using LLMs for stuff in general.

My main reason for building this was I chucking everything into copilot but it gets fuzzy after a couple of weeks... it "forgets" the earlier stuff as it doesn't have an infinite context window.

This is very much just a conceptual thing and I don't know if I'll ever make it useable for other people BECAUSE of the risks you've mentioned... even having it pooping out "this is not medical advice, consult your physician" isn't enough... companies HAVE been sued because people followed something an LLM spat out even though it was dumb.