# Injury Causation Analysis Software > Injury causation analysis software tags every visit and charge related, unrelated, or disputed to the covered incident — cited, with adjudicator sign-off. Canonical page: https://medrecords.ai/product/treatment-relatedness-compensability/ --- NewWe now read the actual MRI and CT imaging — not just the radiologist's 1-page report. [See how →](https://medrecords.ai/product/imaging/) [Home](https://medrecords.ai/) ›[Product](https://medrecords.ai/product/) › — Treatment Relatedness & Compensability Signals Injury Causation Analysis Software ## Which treatment belongs to this injury? Tagged line by line — you adjudicate. Treatment relatedness and compensability software tags every visit and charge related, unrelated, or disputed relative to the covered incident, with the reasoning and citations behind each tag. Tags are provisional signals, not compensability determinations — any tag can be overridden, and a mandatory adjudicator sign-off is required before anything is final. Tags are provisional signals, not a compensability decision. Any tag can be overridden, and nothing downstream — payment, denial, reserve — moves until the adjudicator signs off. [Test a file ](https://medrecords.ai/test-a-file/?src=product-treatment-relatedness-compensability) [Book a demo](https://medrecords.ai/demo/) Adams, Timothy — right knee · Case #IME-4812 Relatedness Line item — Tag — Cite Orthopedic visit · right knee RELATED p.212 PT session · right knee RELATED p.228 Chiropractic · low back UNRELATED p.241 MRI · right knee RELATED p.250 Injection · right knee DISPUTED p.301 Adjudicator sign-off · pending nothing final until signed Line by line Every visit and charge tagged against the covered incident Cited Reasoning and page-level citations behind each tag Human-gated Mandatory adjudicator sign-off before anything moves ### Every tag arrives with its reasoning attached. Each visit and charge is read against the covered incident — mechanism, body part, date — and tagged related, unrelated, or disputed. Open any tag and the reasoning is there: the passages it rests on, cited to the page, 1 click from the source. 3 tags, no false confidence: conflicts go to disputed Every citation clickable to its source page Injection · right knee DISPUTED Why this tag Treating note ties current right-knee symptoms to the covered incident. p.212 Prior injury to the same knee documented in an earlier packet — County Medical, Aug 2020. p.140 Conflicting evidence on causation → routed to the adjudicator, not auto-resolved. Flagged, **not guessed** — the conflict is surfaced for a human read. Before the incident · Prior right-knee complaint · County Medical · Discharge summary · earlier packet · Same body part → apportionment question After the incident · 7 visits, documented improving since 4/02 · Orthopedic visits · PT · MRI · right knee · Each line cited to its source page Both histories on the table — the apportionment call stays yours. ### Pre-existing conditions, separated — not smoothed over. Prior treatment to the same body part is pulled from the record itself and set beside the post-incident care. Lines tied to a pre-existing condition tag unrelated or disputed with the prior-history citation attached — the evidence your apportionment review needs, without the AI making the apportionment call. Same-body-part prior history flagged for apportionment review No percentage invented — evidence surfaced, decision yours ### The sign-off gate: AI tags, the adjudicator decides. Override any tag, one line at a time — the original suggestion and your override are both kept, logged with user and timestamp. Until the adjudicator signs off, the file stays provisional: no payment posts, no denial issues, no reserve moves. The gate cannot be disabled. Per-line overrides, logged and reversible before sign-off AI drafts the signals; the human decision is the record Adjudicator review Case #IME-4812 Chiropractic · low back UNRELATED → DISPUTED Overridden by adjudicator · logged with user + timestamp Remaining tags accepted as suggested AI suggestion and human decision both preserved in the file Sign off & finalize unlocks downstream steps Until signed: **no payment, no denial, no reserve change** . Reasoning trail · injection, right knee preserved in file Signal: treating note links symptoms to covered incident p.212 Counter-signal: prior injury, same knee, earlier packet p.140 Tag: DISPUTED — routed to adjudicator Decision: adjudicator's — recorded beside the AI's suggestion The boundary ### Signals with citations — never a compensability determination. Every tag traces to the pages it rests on, and every conflict is flagged rather than resolved by the machine. That is what makes the output audit-grade and legally defensible: when the determination is questioned later, the file shows the evidence, the AI's signal, and the human decision — separately. [See Verifiable AI Citations ](https://medrecords.ai/product/citations/) ### From claim file to signed relatedness review. 3 steps — the AI does the reading, your adjudicator does the deciding. 01 Upload the claim file Records, bills, and the incident description — any format, any volume, no manual sorting. 02 Every line tagged & cited Related, unrelated, or disputed per visit and charge — reasoning and page citations attached to each. 03 Review, override, sign off The adjudicator accepts or overrides each tag, then signs — only then does anything move downstream. ### Who works the relatedness question. Same tags, different desks: the adjudicator's queue, the TPA's client file, the defense attorney's causation argument. [ Insurance carriers Adjusters open the file with every line pre-read and cited — and keep full authority over the determination. For carriers ](https://medrecords.ai/solutions/insurance-carriers/) [ TPAs & case management Relatedness worked consistently across adjusters and clients, with the reasoning trail ready for any client audit. For TPAs ](https://medrecords.ai/solutions/tpas/) [ Law firms The causation argument, mapped to the record line by line — with the pre-existing history already on the table. For law firms ](https://medrecords.ai/solutions/law-firms/) FAQ ### Relatedness signals, answered. Prior treatment is surfaced from the record itself — earlier packets, earlier providers — and lines tied to a pre-existing condition are tagged unrelated or disputed with the prior-history citation attached. That gives your apportionment review its evidence; the apportionment call itself stays with your team. Yes — any tag, one line at a time, without disturbing the rest. Every override is logged with user and timestamp, and the original AI tag and reasoning stay in the file so the record shows both what was suggested and what was decided. No. Tags are provisional signals with citations, not a compensability determination. Nothing downstream — payment, denial, reserve — moves until the adjudicator reviews and signs off, and the sign-off gate cannot be disabled. Each visit and charge is read against the covered incident — mechanism, body part, and date — and assigned one of 3 tags. Clean matches tag related, clearly separate care tags unrelated, and anything with conflicting evidence, such as a pre-existing complaint in the same body part, tags disputed for human attention. The cited record passages the tag rests on, the linkage (or gap) to the covered incident, and any prior-history references — each citation clickable to its source page. The trail is preserved in the file, so a later reviewer or auditor sees exactly why each line carries its tag. ### Related capabilities [ Condition Progression Tracking How the condition actually moved over time — the context behind each tag. Explore ](https://medrecords.ai/product/condition-progression-tracking/) [ Work Limitation Extraction Restrictions and work-status changes pulled from the same record. Explore ](https://medrecords.ai/product/work-limitation-extraction/) [ Billed-Charge Benchmarking Once relatedness is decided, check whether the charge itself is reasonable. Explore ](https://medrecords.ai/product/medical-bill-review/) ### See your next claim file tagged line by line. Upload a file and get the relatedness tags back, cited — with the sign-off gate exactly where it belongs. Handled under our BAA; never used to train a model. [Test a file ](https://medrecords.ai/test-a-file/?src=product-treatment-relatedness-compensability) [Book a demo](https://medrecords.ai/demo/) [All capabilities →](https://medrecords.ai/product/)