Analysis version v0.4.1 · evidence current through 2026-08-10 · Changelog
Claude, Codex, and Grok each worked the medical record independently, then each attacked the others' conclusions. A fourth build, Claude Fable, later re-derived the analysis blind from the compiled evidence — then attacked its own result and the published one alike. What survived is published here with its counterevidence, its gaps, and every correction logged.No diagnosis is offered. This is a research record built for clinicians to challenge.
Multi-model agreement is not validation. No licensed clinician has reviewed these pages. Where the models disagreed, both positions stay on the record.How this works →
Initial research preview — incomplete and open to correction.
Not medical advice, a diagnosis, or a treatment plan. Drift0r granted permission to publish this research; permission is not endorsement. Full limits and disclosures ↓
Plain-language orientation. Every point below traces to a row in the claim record; where something is the patient's own account rather than an instrument's, it says so.
Working research models, not diagnoses. Each is published with a count of what it accounts for and what it fails to explain.
Rather than one rare disease explaining everything, the record may reflect several distinct problems — bone, metabolism, nerves, and immune system — happening at once. This is a way of organizing the findings, not a diagnosis.
Fits 5 documented findingsDoesn't account for 4
H1 · Layered skeletal–metabolic–neurologic–immune stack (infection detachable)
Full record, support, and counterevidence →A single gene affecting how bone is built could account for bone loss this marked at this age. The genetic panel that would test this has not been done, and a negative panel would not fully close the question.
Fits 3Doesn't account for 4
H2 · Monogenic / constitutional early-onset osteoporosis spectrum
Full record →Urine calcium has been collected three times with mixed results — two of the three below the usual threshold — and never under the controlled conditions needed to read it. The markers that show whether new bone is being built do not appear anywhere in the test record, though that is not proof they were never drawn. This is a gap in the documented workup rather than an established finding. A kidney condition called distal renal tubular acidosis, which can produce this same combination of alkaline urine, urinary calcium loss and weakened bone, was considered here: two of its expected signatures are absent from his results, so it is not favoured — though only a clinician can judge whether its milder, incomplete form deserves a formal test.
Fits 3Doesn't account for 2
H3 · Calcium-flux / incomplete bone-turnover phenotyping module
Full record →A Bartonella-type infection could contribute to the joint and bone-surface findings. Specialty lab results are disputed and standard testing was negative, so this stays low confidence and separable from the rest of the picture.
Fits 2Doesn't account for 4
H4 · Bartonella-spectrum residual contributor (specialty LDT contested)
Full record →One antibody subclass is persistently low and another persistently high, across repeated draws. The pattern repeats; what it means is unresolved. A vaccine-response test would be needed to interpret it.
Fits 2Doesn't account for 2
H5 · Humoral abnormality / possible specific antibody deficiency
Full record →The baseline possibility that several ordinary problems coexist without one rare disease tying them together. Kept on the record as a discipline against forcing the findings into a single story.
H-NULL · Multi-independent axes (null model / no single unifier)
Full record →All 6 working models with their support and counterevidence →
Bands from documented findings and patient-reported history through open questions. Hypothesis-kind claims never appear as documented findings. Every node is a real record identifier (CLM, H, lit, UQ). Connectors are undirected and unweighted; relationship meaning is carried by band headings, a compact legend, accessible node descriptions, and the authoritative table — not by repeating the same verb on every line. The diagram terminates in unknowns — there is no answer band. Nodes are links or keyboard-focusable; CLM, H, UQ, and lit nodes expose an accessible preview of approved public fields. Bands wrap onto more rows rather than crowding their nodes together. A complete table equivalent sits directly below. On narrow screens, swipe hypothesis-centered evidence-path cards.
Swipe evidence-path cards → each card is one hypothesis. Ends in open questions — no diagnosis / answer band.
H1 · architecture
Confidence: medium
+3 more on full record
+2 more on full record
UQ-0003, UQ-0013, UQ-0017
Full H1 record on Working model →
Flow ends at open questions. There is no diagnosis or answer band.
H2 · single primary
Confidence: medium
+1 more on full record
+2 more on full record
UQ-0013, UQ-0005, UQ-0020
Full H2 record on Working model →
Flow ends at open questions. There is no diagnosis or answer band.
H3 · module
Confidence: medium
+1 more on full record
UQ-0002, UQ-0005
Full H3 record on Working model →
Flow ends at open questions. There is no diagnosis or answer band.
H4 · residual
Confidence: low
+2 more on full record
UQ-0017
Full H4 record on Working model →
Flow ends at open questions. There is no diagnosis or answer band.
H5 · module
Confidence: low
UQ-0007
Full H5 record on Working model →
Flow ends at open questions. There is no diagnosis or answer band.
H-NULL · null model
Confidence: medium
— none listed —
— none listed —
UQ-0017
Full H-NULL record on Working model →
Flow ends at open questions. There is no diagnosis or answer band.
| Hypothesis | Kind | Confidence (word) | Would explain (claims) | Does not explain (claims) | Supporting lit | Contradicting lit | Open questions |
|---|---|---|---|---|---|---|---|
| H-NULL Multi-independent axes (null model / no single unifier) | null_model | medium | — | — | +0 | −0 | UQ-0017 |
| H1 Layered skeletal–metabolic–neurologic–immune stack (infection detachable) | architecture | medium | CLM-0003, CLM-0015, CLM-0023, CLM-0027, CLM-0013 | CLM-0030, CLM-0031, CLM-0036, CLM-0018 | +3: lit-0015, lit-0021, lit-0205 | −0 | UQ-0003, UQ-0013, UQ-0017 |
| H2 Monogenic / constitutional early-onset osteoporosis spectrum | single_primary | medium | CLM-0003, CLM-0005, CLM-0006 | CLM-0023, CLM-0027, CLM-0036, CLM-0112 | +5: lit-0287, lit-0290, lit-0294, lit-0293, lit-0015 | −3: lit-0340, lit-0341, lit-0134 | UQ-0013, UQ-0005, UQ-0020 |
| H3 Calcium-flux / incomplete bone-turnover phenotyping module | module | medium | CLM-0015, CLM-0010, CLM-0011 | CLM-0023, CLM-0027 | +4: lit-0013, lit-0152, lit-0014, lit-0153 | −3: lit-0337, lit-0338, lit-0339 | UQ-0002, UQ-0005 |
| H4 Bartonella-spectrum residual contributor (specialty LDT contested) | residual | low | CLM-0037, CLM-0040 | CLM-0003, CLM-0018, CLM-0015, CLM-0039 | +2: lit-0037, lit-0050 | −4: lit-0049, lit-0042, lit-0047, lit-0055 | UQ-0017 |
| H5 Humoral abnormality / possible specific antibody deficiency | module | low | CLM-0027, CLM-0034 | CLM-0003, CLM-0023 | +3: lit-0205, lit-0089, lit-0096 | −1: lit-0206 | UQ-0007 |
A negative test is not the same as an impossibility. These entries record what was run and what it showed — they are not clinical exclusions, and no licensed clinician has reviewed them.
| Suggestion | What was run | Result | Still open |
|---|---|---|---|
| Celiac diseaseCLM-0078 | Antibody panel (tTG, DGP, EMA) plus total IgA2025-10 | All negative. Total IgA sufficient, so the usual false-negative route does not apply. | No duodenal biopsy and no HLA-DQ2/DQ8 typing. Gluten intake before the draw is not documented, and a negative result on a reduced-gluten diet is hard to interpret. |
| Lyme diseaseCLM-0038 | Multi-method serology and PCR across several labs2023 onward | Negative by every method documented in the summaries, on both specialty and CDC/NYS criteria. | — |
| Mold illness / CIRSCLM-0052 | Mold toxicity and allergy testing, plus household mold removal~2026 | Reported by the patient as negative, with no clear symptom change after the mold was removed from the home. The underlying lab reports are not in the public record. | An extended away-from-home environmental trial has been discussed but not carried out. |
| Mast cell activation / mastocytosisCLM-0030 · CLM-0031 · CLM-0032 · CLM-0065 | KIT mutation testing, serial serum tryptase, urine mast-cell mediators2025 – 2026 | Mixed, and not a clean negative. Tryptase was above the reference range on both documented draws; the record attributes that to hereditary alpha-tryptasemia, a genetic trait the patient reports testing positive for. KIT was negative and urine mediators were normal. | The KIT assay's specimen, method, and limit of detection are not documented, so systemic mastocytosis is not fully excluded — this is on the project's own list of things a clinician should not miss. Bone marrow biopsy is deferred until bone density makes the procedure safer. |
| Lupus, rheumatoid arthritis, Sjögren's, sclerodermaCLM-0044 | Full autoantibody panel (ANA, dsDNA, ENA, C3/C4), rheumatoid factor, anti-CCP2021 – 2026 | Unrevealing on every panel run. Inflammatory markers were normal on the three documented CRP draws and the three documented ESR draws. | No CRP or ESR value after 2025-08-05 appears in the summaries, so the inflammatory picture is not current. |
| Ehlers-Danlos syndromeCLM-0081 | Genetic and clinical assessmentreported history | Reported by the patient as previously cleared on both genetic and clinical grounds. | Patient-reported only — no assessment document is in the public record, and which genes were on the panel is not documented. |
| Thyroid diseaseCLM-0079 | Thyroid panel with antibodies (TSH, free T4, T3, reverse T3, TPO, thyroglobulin), repeated2021 – 2025 | Euthyroid across every draw; antibodies negative. | One isolated uptake value sat slightly outside its reference range in 2022, which carries little weight on its own. No thyroid panel after 2025 appears in the summaries. |
| Parathyroid diseaseCLM-0017 | Intact PTH with serum calcium and vitamin D, repeated2021 – 2026 | Calcium normal, PTH low-normal — the opposite of the expected pattern. | — |
| Heavy metal toxicityCLM-0080 | Heavy metals panel (arsenic, cadmium, cobalt, lead, mercury)2026-06 | All below reporting cutoffs or undetectable. | — |
| Myeloma or a blood-protein disorderCLM-0028 | Serum protein electrophoresis and free light chains2022 and 2026 | Polyclonal, no abnormal clone, normal light-chain ratio. | — |
| COVID-19 vaccine injuryCLM-0051 | Symptom-onset timeline against vaccination datestimeline | Onset is reported as around 2017, several years before COVID-19 vaccines existed. | The onset date comes from a patient-compiled timeline rather than a dated clinical record. |
| It is psychological / somatic symptom disorderCLM-0046 · CLM-0047 · CLM-0048 | Psychiatric and psychological evaluation history, including MMPI-2-RFthrough 2025 | A somatic symptom disorder label applied in 2021 was formally reversed in 2025, and the 2025 MMPI-2-RF found no evidence of over-reporting of somatic symptoms. The bone density, imaging, and laboratory findings are objective and instrument-recorded. | The evaluation history other than the MMPI-2-RF result is patient-reported. What the record supports is the reversal of that specific label — not a finding that no psychiatric factor plays any part. Treating the medical findings as primary is this project's research position, not a clinical adjudication. |
| Cushing's syndrome / endogenous hypercortisolismCLM-0106 · CLM-0003 · CLM-0020 · CLM-0022 | Morning serum cortisol (twice), 24-hour urine free cortisol, plasma ACTH (twice)2021 – 2025 | All within reference — morning cortisol 19.7 and 20.4 µg/dL (ref 4.0–22.0), urine free cortisol 22.8 µg/24h (ref ≤60), ACTH 19.4 and 11 pg/mL (ref 6–50). The endocrine summary's own reading is "no Cushing's or adrenal insufficiency signal." | Both morning cortisol values sit high in their interval — 19.7 and 20.4 against a ceiling of 22.0. Two morning draws and one urine collection do not exclude cyclic or mild autonomous cortisol excess — the screens directed at that (overnight dexamethasone suppression, late-night salivary cortisol) appear nowhere in the record. That absence matters here because the record also carries severe lumbar-spine bone loss, an off-therapy hormone pattern its own summaries read as central, and a small 2021 pituitary enhancement focus not seen on follow-up — the combination those screens exist to interrogate. No tally of past prescribed glucocorticoid exposure appears in the summaries either. |
| Hereditary hemochromatosis / iron overloadCLM-0107 · CLM-0108 | Full iron studies on three dated draws, ferritin serially through mid-20262021 – 2026 | Read by the endocrine summary as iron-replete, with the rheumatology summary calling iron indices otherwise unremarkable. Transferrin saturation 35% and 43% (ref 20–48) and 38% (ref 15–55) — never flagged high. Ferritin ran high-normal to transiently high — 398 ng/mL against one table's 38–380 interval, then 466 in August 2025 and 450 in January 2026 against ~30–400, settling to 294 by mid-2026 — with the rheumatology summary noting the upper limit differed by laboratory (400 vs 380). TIBC sat below its 250–450 interval on all three dated panels, which the endocrine summary reads as fitting an iron-replete or inflammatory pattern, not deficiency. | No HFE genotyping appears in the record. The 2025 ferritin rise has no stated cause; ferritin climbs with inflammation as readily as with iron loading, and the concurrent normal saturation argues against loading without settling what it was. |
| Myasthenia gravis (for the speech and weakness symptoms)CLM-0109 · CLM-0111 | Acetylcholine-receptor antibody2022-09 | <0.30 nmol/L against a reference of ≤0.30 — negative. | A negative acetylcholine-receptor antibody bears on classic seropositive myasthenia only — MuSK and LRP4 antibodies and single-fiber EMG appear nowhere in the record, and the draw predates the 2026 speech symptom by roughly four years. On that symptom the record holds only the patient's own video statements, which pull in different directions: he attributes it to his jaw joint, and near the end of the same recording observes his speech getting less clear as he tires. Whether it is formally fatigable — worse with minutes of use, recovering with rest — is the zero-cost observation no summary documents, and it is what separates neuromuscular-junction territory from the mechanical explanation. |
One specialty laboratory returned a positive signal. Independent testing did not confirm it. Both results stay on the record side by side and are never merged into a single answer — the same provenance rule applied to DXA and every other value on this site.
Related claims: CLM-0036. Channels cannot be merged into a single positive.
Related claims: CLM-0037. Channels cannot be merged into a single positive.
The method is auditable. The conclusions are not established. Those are different things, so they are listed separately.
Drift0r directly granted permission to publish this repository, website, and approved data. That permission does not imply that Drift0r or any clinician endorses the findings.