How this watch works

Home · Topics · Latest · Calendar · Browse tags

Research scope, evidence rules and operating instructions. The reading home is the cover page. The schedule remains Tuesday and Friday at 9 a.m. Eastern, with meaningful-change notifications.

Agreed scope

Check Tuesday and Friday at 9 a.m., America/New_York.
Follow human studies worldwide, including early-phase results and credible announcements of planned or newly registered human trials. Identify planned, registered, recruiting, and completed studies separately.
Highlight U.S. regulatory status and availability separately from worldwide development.
Include efficacy, meaningful safety findings, negative trials, clinical holds, discontinuations, new indications, new drugs, and screening or therapy announcements within the active topics.
Include aesthetic and dental drugs, devices and clinician-delivered procedures studied in humans, including dental alignment, retainers, cellulite and biological hair-pigment restoration.
Sleep therapies explicitly include insomnia drugs, cognitive behavioral therapy for insomnia (CBT-I), validated digital therapeutics, circadian-rhythm treatments, narcolepsy and idiopathic hypersomnia, restless legs syndrome, apnea drugs/devices and clinically useful detection. Keep each disorder and indication separate; discover new human-stage entrants beyond the named programs.
Focus on formal academic or regulated clinical development. Exclude gray-market products, unsupervised protocols, and routine animal or laboratory-only stories.
Include relevant rumors about these programs or potential new human-stage therapies, even before official confirmation or registration. Reddit posts/comments, podcast and talk transcripts, Q&A remarks, interviews and other informal sources are in scope. The formal-development focus limits the subject matter, not the sources allowed for a labeled rumor.
Notify only when there is meaningful new evidence, a material development milestone, a noteworthy rumor or change in its status, a monitoring failure that needs attention, or a required user decision. Keep unchanged checks quiet.

Rumors and informal signals

Keep these within the relevant existing topic, in a clearly labeled Rumors and informal signals section when there is something to record. Do not create another category or add empty rumor sections to every topic.

Label each item Rumor — unconfirmed, Informal statement — unconfirmed, or Speculation/inference as appropriate. A verified recording proves what a speaker said, not that the predicted trial result or milestone will occur. Distinguish the speaker's claim from the assistant's interpretation.
Include specific, relevant leads about readouts, timing, new trials, indications, safety or development decisions. A single anonymous post or offhand remark can be included; official confirmation or independent corroboration is not required for inclusion. Explain the available basis without assigning invented confidence percentages.
Cite the original post/comment permalink, or episode/talk title, speaker, date and timestamp/transcript passage when available. Preserve enough context to avoid turning a hypothetical remark into an announcement. If only a repost or secondhand account is accessible, cite it as such. If the original source cannot be found, say so; never invent a link, quotation or attribution.
Look for corroboration or contradictory evidence where practical. Reposts, upvotes and multiple accounts repeating one source are not independent confirmation. Note anonymity or an apparent promotional interest when relevant to interpreting the claim.
Keep efficacy, safety, approval and timing claims explicitly unverified until evidence supports the particular claim. Do not fold rumored results into the evidence summary or turn rumored dates into announced dates.
Track the source/publication date separately from the alleged event date. Use a stable rumor key and a dated status history: unconfirmed, corroborated but unconfirmed, confirmed, contradicted, retracted, or stale/unresolved. Link subsequent evidence and preserve the original attribution. An elapsed forecast becomes stale, not automatically disproved.
Seek updates to unresolved rumors during later checks. Announce material confirmation, contradiction or retraction, while avoiding repeated alerts for the same unsupported claim. A later real trial result is a new evidence event even if a rumor anticipated it.

Rumor entry format: Label and claim; program/topic; who said it and where; source date; alleged timing; direct source or access limitation; corroboration/contradiction; why it may matter; next check; status history. Omit unavailable details rather than guessing.

How each check works

Read this file, README.md, WATCHLIST.md, catalog.json, the active topic notes, LATEST.md, MILESTONES.md, and state.json. Use the current files as the source of truth for scope. Track visible active condition topics; a record with role: "hub" retains historical coverage and links but is not a duplicate monitoring assignment. Read its linked inherited evidence as needed.
For a topic without a completed baseline, establish the major active programs, trial identifiers, current status, latest material evidence, and announced next milestones from primary sources. Search at least the preceding 90 days and older pivotal context as needed. Label older context as baseline, not breaking news.
For subsequent checks, search from that topic's last successful check with a seven-day overlap to catch indexing delays. Compare event dates and publication dates; inspect original dated sources instead of treating a recent repost as a new result.
Check all active topics using trial registries, primary papers, regulator announcements, official conference abstracts, trial investigator sites, and sponsor releases. Also search relevant Reddit discussions, podcast/talk transcripts and informal remarks for rumors, following the rules above. Search across categories for newly named entrants as well as following existing names. Use primary sources to substantiate established scientific claims; informal sources can substantiate that a rumor or comment exists without proving its substance.
Label evidence as a peer-reviewed human study, conference abstract, preprint, sponsor topline announcement, trial-registry update, regulatory action, unconfirmed rumor/informal statement, or speculation/inference. A sponsor's clearance to run a trial is not marketing approval; an estimated or rumored readout is not a completed result.
For meaningful findings, record the event and source dates, drug/test and trial ID, phase/design, population, sample size, comparator, follow-up, primary endpoint, magnitude and uncertainty where available, key safety findings, limitations, and why the update matters. Use direct links. Do not manufacture unavailable numbers.
For aesthetic and dental topics, distinguish visible improvement from collagen/pigment biomarkers; laxity from volume loss; cellulite dimpling from fat volume; scar prevention from erasing established scars; congenital tooth agenesis from acquired tooth loss; and early enamel remineralization from replacing missing tooth structure. For orthodontics, separate fewer visits and faster manufacturing from shorter total treatment, final bite quality, root/gum safety and long-term retention. Keep cartilage structure distinct from pain/function and avoided joint replacement; apnea indices from cardiovascular outcomes; sleep onset/maintenance and device sleep-stage estimates from insomnia remission, daytime function and durable benefit; orexin antagonists for insomnia from orexin agonists for hypersomnolence; remission from cure; sensory test gains from everyday function; and hair counts from visible regrowth and durable benefit. Keep screening accuracy, detection yield, stage shift, and demonstrated health benefit distinct. Keep biomarker changes distinct from clinical outcomes; keep measured lean mass distinct from strength and function. Report adverse and null findings with the same care as positive findings.
Append dated findings to the affected topic pages and update their current summaries without erasing historical notes. Keep rumors visibly separate within the topic. Refresh LANDSCAPE.md, the affected topic’s Current state and milestone preview, the dated MILESTONES.md calendar, LATEST.md and README.md previews when material facts change, preserving the distinction between announced dates, sponsor forecasts, registry estimates, analyst estimates and explicitly unconfirmed rumored dates. Keep INITIAL_BRIEF_2026-09-12.md as a dated historical snapshot. Write a dated run record under updates with per-topic coverage, sources checked, findings or no material change, and access gaps. Include informal-source coverage and unresolved rumor follow-up when applicable.
Update state only after reading back the written files. Advance a topic's successful-check timestamp only if its research was completed. A blocked source or incomplete topic is an explicit coverage gap, not evidence of no news. Retry incomplete coverage on the next check.
Deduplicate by study/program, event date, endpoint or milestone, source and evidence/status type. A later full paper or longer follow-up is a new event when it adds material information; a repeated press story is not. Record rumor keys and status transitions in state.json's seen_events alongside the linked topic entry; keep a confirming research event distinct from the earlier rumor. Track the original source chain so repeated rumors do not appear independently corroborated.
For a meaningful update, post a concise digest in this task: what changed, how strong the evidence is, why it matters, the next milestone if known, and source links. Put rumor items under an explicit rumor label within the relevant topic's digest, with attribution and uncertainty. Link the updated local notes. Group overlapping news once and cross-link its topic pages. Do not issue routine no-news notifications.

Update entry format

Use a dated entry with: What changed; evidence type; study and population; result; limitations and safety; U.S. status; next milestone; sources; interpretation. Omit inapplicable fields rather than filling them with guesses.

This notebook tracks research developments.

Setup status

All twenty-five active topics have a selective initial baseline as of September 12, 2026, including expected-update timing or an explicit statement that no reliable date is known. The same-day expansion adds joint preservation, sleep, liver health, cancer vaccines, autoimmune therapies, non-opioid pain, kidney preservation, hearing/vision and aesthetic hair-loss research. Organ-specific cancer and cardiovascular imaging screening are included within their existing topics. A further same-day expansion adds six aesthetic/dental categories, with explicit dental-alignment coverage, plus cellulite and hair-graying subtopics. Subsequent checks compare against this baseline and discover new human-stage programs within the saved scope.

Automation registered and active on September 12, 2026: therapy-and-screening-watch (Therapy and screening watch), returning to this task on Tuesday and Friday at 9 a.m. Eastern. Manage scheduling through the Codex scheduled-task controls; editing this paragraph alone does not change the schedule.

The scheduled task needs this local directory to remain available, with the computer on and the desktop app running. See official scheduled-task documentation.

Maintaining the library

Sleep organization: Since October 3, 2026, the sleep-health group contains five condition topics: sleep_insomnia, sleep_apnea, sleep_hypersomnolence, sleep_circadian and sleep_restless_legs. The stable sleep ID/path remains an active history hub with role: "hub" and browseHidden: true; omit it from topic-card counts and duplicate monitoring. Prior feed links and note anchors remain in that hub. The new pages inherit existing evidence, not completed fresh checks; finish their focused baselines before advancing successful-check timestamps. Record a cross-condition event once and link the other page.
Topic notes are the evidence source. Keep the existing filenames and stable IDs. The watchlist states scope; catalog.json records groups, tags and topic paths. A primary group is navigation, not a medical classification or priority ranking.
One event, multiple links. Preserve stable event keys and status history in state.json. Cross-link overlapping topic pages; show the event once in a digest. Do not turn a reorganization, repost or repeated rumor into a new finding.
Common topic order: Current state; Questions to follow; Programs and next milestones; Rumors and informal signals only when an actual item exists; Dated history. Preserve source dates, caveats, negative findings, quotations. Keep important older heading anchors when moving content.
Separate program fields: program and indication/population; development stage; evidence type and limitations; U.S. approval/availability with exact formulation/indication; next milestone with timing/source. Do not infer evidence strength from phase or equate a trial authorization with marketing approval. Existing detailed narrative evidence remains valid; use the topic template when adding or substantially updating a program.
Views: LANDSCAPE.md carries current summaries for every topic; MILESTONES.md orders expected events with date precision; LATEST.md contains a concise dated feed; BROWSE.md offers cross-cutting tags. README.md has a short preview and grouped navigation. Update affected views together after a meaningful result or milestone change.
Precise full-note links: Every LATEST.md entry links to the specific finding with [full notes](topics/<file>.md#<passage-id>). Reuse its heading ID, or add a stable <span class="note-anchor" id="note-YYYY-MM-DD-short-description"></span> immediately before the relevant paragraph. Preserve that ID if the finding moves into Dated history. A broad section or the top of a topic is not a sufficient destination. Navigation-only edits do not advance research timestamps or create news alerts.
Calendar: use exact days only where supported; retain month, quarter and year windows. Keep scheduled events, sponsor forecasts, registry estimates, regulatory requirements and unconfirmed rumors distinguishable. When a window passes, investigate and mark its status rather than silently treating the milestone as completed. Keep undated and unresolved programs visible.
Archive: INITIAL_BRIEF_2026-09-12.md remains the fixed historical snapshot. Preserve existing update logs; add dated records and maintain updates/README.md. Organization-only work must not advance any topic’s last_successful_check_utc or the global research-run timestamp.
New scope: add a stable ID and path to catalog.json and state.json, a grouped watchlist entry, appropriate tags, and a topic page; update all affected indexes. Discovering a program within an existing topic need not create another category. Pausing requires explicit user intent, not a low volume of news.
Validation: compare all current IDs/paths, check local files and anchors, retain every prior event, and verify dates/evidence/status qualifiers. Read back state and the saved scheduled-task configuration after changes.

Scope decisions

"Galeri" refers to Galleri, the GRAIL multi-cancer detection test.
Localized fat removal includes both regional subcutaneous fat and visceral fat. Follow notable formal human drug programs, including tesamorelin's possible new indications; distinguish reductions in these depots from liver-fat reduction and overall weight loss.
Screening devices and tests are included even when their studies do not use drug-trial phase terminology.
The research library added psilocybin and related drugs on September 12, 2026. Track named products and classes separately, including aliases after program/company renaming. Distinguish FDA drug approval, investigational studies, and state-regulated services.
SSRIs and newer antidepressants are a separate active category. Cross-link psychedelic comparisons and combination studies without merging the drug classes or reporting the same event twice.
The research library added all eight recommended additions, both screening expansions and hair loss on September 12, 2026. The two screening expansions live within the existing cancer-screening and cardiovascular topics; hair loss is its own aesthetic category.
Cross-link overlapping liver, kidney and sleep findings with GLP-1/CV topics, joint findings with muscle/pain topics, and autoimmune alopecia with immune therapies. Keep one event record/digest item linked to every relevant topic. A new category establishes research scope, not a personal treatment or screening recommendation.
The research library added six further aesthetic categories on September 12, 2026: skin aging/laxity, scars/stretch marks, facial volume, pigmentation, acne/rosacea and dental research. Dental alignment is explicitly included with regeneration/repair. Cellulite and hair graying are subtopics in the existing body-contour and hair pages.
Aesthetic and dental scope includes formal human studies of drugs, devices and clinician-delivered procedures. Track visible and functional outcomes, durability, adverse effects and U.S. availability; distinguish biochemical changes and simulations from observed human benefit. Cross-link shared interventions without duplicating events.
Sleep therapies were added to the research scope on September 12, 2026. Broaden the existing stable sleep topic rather than duplicate it; retain 25 topics and all original apnea/narcolepsy notes. The expanded sleep baseline is selective and does not imply a completed all-topic monitoring run.
September 13, 2026 scope expansion: retain the stable cancer_vaccines, joint_preservation and ssri_antidepressants topics while broadening them to cancer treatment, bone health and mental-health treatments beyond depression. Add allergy_airway, infectious_disease and genetic_screening, bringing the library to 28 active topics. Future antibiotics, resistance, bacteriophages, vaccines, antivirals and long-acting infection prevention are explicitly included.
Genetics coverage: include inherited pathogenic variants/cascade testing, chronic-disease PRS, reproductive carrier screening, PGT-M/PGT-SR/PGT-A, PGT-P research, pharmacogenomics and newborn sequencing. Separate analytic accuracy, disease association, risk-model validation, changes in management and measured clinical outcomes. Distinguish modeled embryo risk reduction from prospective offspring outcomes; retain ancestry, absolute-risk and embryo-number limitations. Do not confuse embryo selection with DNA editing or commercial availability with validated clinical benefit.
Expanded outcomes: separate bone density from fractures; cancer response from progression-free or overall survival; and psychiatric symptom change from remission, function, relapse and safety. Do not extrapolate one psychiatric diagnosis or cancer setting to another. Cross-link related studies once. The scope baseline does not refresh untouched historical evidence or constitute a complete monitoring/rumor run.