find-cohort-gap

Name: find-cohort-gap
Author: Aperivue

43installs53forks

How to install find-cohort-gap?

aperivue/medsci-skills/find-cohort-gap

$npx -y skills add Aperivue/medsci-skills --skill find-cohort-gap

Installs into the current project.

Summary

Research gap finder for longitudinal cohort databases. Profiles cohort strengths, matches PI expertise, scans literature saturation, and outputs ranked topic proposals with gap evidence. Works with any cohort: NHIS, UK Biobank, institutional EMR, health checkup registries, or dis

Files · 1

View on GitHub

SKILL.md

1	# Find-Cohort-Gap Skill
2
3	You are assisting a medical researcher in systematically discovering novel, publishable
4	research topics from a cohort database. Your approach combines cohort variable profiling,
5	PI expertise matching, literature saturation scanning, and multi-pattern gap scoring to
6	produce ranked topic proposals with evidence of novelty.
7
8	This skill fills a gap that no existing tool addresses: **DB variables -> literature gap
9	-> research question**. Existing tools (PICO, FINER, SciSpace, Elicit) work from
10	literature to gaps. This skill works from the data outward.
11
12	## Communication Rules
13
14	- Communicate with the user in their preferred language.
15	- All literature citations, variable names, and medical terminology in English.
16	- Be direct about weak topics — kill early, save time.
17
18	## Key Directories
19
20	- Output: User-specified directory (default: current working directory)
21	- References: `${CLAUDE_SKILL_DIR}/references/` for templates and rubrics
22
23	---
24
25	## Phase 0: Cohort Intake
26
27	The cohort does not have to be one this skill has heard of. Route on what the user
28	actually has.
29
30	\| The user has… \| Do this \|
31	\|---------------\|---------\|
32	\| A named public cohort (NHIS, UK Biobank, KNHANES, …) \| Fill the profile from published documentation. Cite the source for every field. \|
33	\| A codebook / data dictionary / CSV export of their own registry or EMR extract \| Run the input adapter below. This is the common case — an institutional registry or single-centre export that no public documentation describes. \|
34	\| A review, guideline, or preprint defining the clinical domain \| Attach it as domain context (`--context`), as a file or a URL. \|
35
36	### Input adapter (local codebook / documents)
37
38	```bash
39	python3 "${CLAUDE_SKILL_DIR}/scripts/build_cohort_profile.py" \
40	--codebook data_dictionary.csv \
41	--context narrative_review.pdf --context https://example.org/guideline \
42	--cohort-name "Institutional CT registry" --out-dir .
43	```
44
45	Formats: `.csv` / `.tsv` / `.json` / `.md` / `.txt` (stdlib), `.xlsx` (needs `openpyxl`),
46	`.pdf` (needs `pdftotext`). A `.csv` is auto-detected as a codebook (rows are
47	variables) or a data export (the header row is the variable list). Writes
48	`cohort_profile.md` + `cohort_profile.json` (+ `context_extract.md`).
49
50	Do not read the codebook yourself and summarise it. Paraphrasing a variable name,
51	merging two that look alike, or inventing one the cohort does not have poisons every
52	downstream claim — the intersection matrix, the feasibility gate, and eventually the
53	manuscript's Methods. The adapter enumerates variables verbatim with provenance
54	(`file:row`) instead, which is the dictionary-first discipline a reviewer expects of a
55	DB-backed study. Read `cohort_profile.md`; do not re-derive it.
56
57	What the adapter infers (and shows its work for): the variable cluster map, **serial
58	/ repeated-measure groups (evidence for P1 Longitudinal Advantage), and endpoint
59	candidates** (evidence for P2 Endpoint Upgrade). Every cluster assignment records the
60	keyword that triggered it, and a variable matching nothing is left `unclassified` rather
61	than forced into a bucket — review those, since the lexicon is not exhaustive.
62
63	### What the adapter cannot know — ASK, never guess
64
65	A codebook lists variables. It does not state any of the following, and each is emitted
66	as `[UNKNOWN - ask the user]`:
67
68	1. Sample size (N at baseline, N with follow-up)
69	2. Time span (enrollment period, follow-up duration, measurement intervals)
70	3. Known limitations (healthy volunteer bias, attrition, missing-data patterns)
71	4. Existing publications from this cohort (to avoid duplicating them)
72	5. IRB status and data-access route
73
74	Collect these from the user before Phase 2. A guessed N does not merely sit there — it
75	flows into the Phase 5 feasibility gate, which then passes (or fails) for a reason that
76	has nothing to do with the cohort.
77
78	Also confirm the setting (institution type, country, population type) and any
79	special strengths the variable names cannot reveal — registry linkage, biobank
80	availability, a distinctive population.
81
82	Gate: Present the cohort profile summary, including the `[UNKNOWN]` list and the
83	unclassified variables. Confirm before proceeding.
84
85	---
86
87	## Phase 1: PI/CA Profiling
88
89	Profile the intended PI or corresponding author to find topic-expertise alignment.
90
91	1. Search PubMed for the PI's recent publications (last 5 years).
92	- Use `/search-lit` E-utilities:

People who liked this also viewed

6 picks

Type

find-cohort-gap

43installs53forks

View on GitHub

How to install find-cohort-gap?

aperivue/medsci-skills/find-cohort-gap

$npx -y skills add Aperivue/medsci-skills --skill find-cohort-gap

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Find-Cohort-Gap Skill
2
3	You are assisting a medical researcher in systematically discovering novel, publishable
4	research topics from a cohort database. Your approach combines cohort variable profiling,
5	PI expertise matching, literature saturation scanning, and multi-pattern gap scoring to
6	produce ranked topic proposals with evidence of novelty.
7
8	This skill fills a gap that no existing tool addresses: **DB variables -> literature gap
9	-> research question**. Existing tools (PICO, FINER, SciSpace, Elicit) work from
10	literature to gaps. This skill works from the data outward.
11
12	## Communication Rules
13
14	- Communicate with the user in their preferred language.
15	- All literature citations, variable names, and medical terminology in English.
16	- Be direct about weak topics — kill early, save time.
17
18	## Key Directories
19
20	- Output: User-specified directory (default: current working directory)
21	- References: `${CLAUDE_SKILL_DIR}/references/` for templates and rubrics
22
23	---
24
25	## Phase 0: Cohort Intake
26
27	The cohort does not have to be one this skill has heard of. Route on what the user
28	actually has.
29
30	\| The user has… \| Do this \|
31	\|---------------\|---------\|
32	\| A named public cohort (NHIS, UK Biobank, KNHANES, …) \| Fill the profile from published documentation. Cite the source for every field. \|
33	\| A codebook / data dictionary / CSV export of their own registry or EMR extract \| Run the input adapter below. This is the common case — an institutional registry or single-centre export that no public documentation describes. \|
34	\| A review, guideline, or preprint defining the clinical domain \| Attach it as domain context (`--context`), as a file or a URL. \|
35
36	### Input adapter (local codebook / documents)
37
38	```bash
39	python3 "${CLAUDE_SKILL_DIR}/scripts/build_cohort_profile.py" \
40	--codebook data_dictionary.csv \
41	--context narrative_review.pdf --context https://example.org/guideline \
42	--cohort-name "Institutional CT registry" --out-dir .
43	```
44
45	Formats: `.csv` / `.tsv` / `.json` / `.md` / `.txt` (stdlib), `.xlsx` (needs `openpyxl`),
46	`.pdf` (needs `pdftotext`). A `.csv` is auto-detected as a codebook (rows are
47	variables) or a data export (the header row is the variable list). Writes
48	`cohort_profile.md` + `cohort_profile.json` (+ `context_extract.md`).
49
50	Do not read the codebook yourself and summarise it. Paraphrasing a variable name,
51	merging two that look alike, or inventing one the cohort does not have poisons every
52	downstream claim — the intersection matrix, the feasibility gate, and eventually the
53	manuscript's Methods. The adapter enumerates variables verbatim with provenance
54	(`file:row`) instead, which is the dictionary-first discipline a reviewer expects of a
55	DB-backed study. Read `cohort_profile.md`; do not re-derive it.
56
57	What the adapter infers (and shows its work for): the variable cluster map, **serial
58	/ repeated-measure groups (evidence for P1 Longitudinal Advantage), and endpoint
59	candidates** (evidence for P2 Endpoint Upgrade). Every cluster assignment records the
60	keyword that triggered it, and a variable matching nothing is left `unclassified` rather
61	than forced into a bucket — review those, since the lexicon is not exhaustive.
62
63	### What the adapter cannot know — ASK, never guess
64
65	A codebook lists variables. It does not state any of the following, and each is emitted
66	as `[UNKNOWN - ask the user]`:
67
68	1. Sample size (N at baseline, N with follow-up)
69	2. Time span (enrollment period, follow-up duration, measurement intervals)
70	3. Known limitations (healthy volunteer bias, attrition, missing-data patterns)
71	4. Existing publications from this cohort (to avoid duplicating them)
72	5. IRB status and data-access route
73
74	Collect these from the user before Phase 2. A guessed N does not merely sit there — it
75	flows into the Phase 5 feasibility gate, which then passes (or fails) for a reason that
76	has nothing to do with the cohort.
77
78	Also confirm the setting (institution type, country, population type) and any
79	special strengths the variable names cannot reveal — registry linkage, biobank
80	availability, a distinctive population.
81
82	Gate: Present the cohort profile summary, including the `[UNKNOWN]` list and the
83	unclassified variables. Confirm before proceeding.
84
85	---
86
87	## Phase 1: PI/CA Profiling
88
89	Profile the intended PI or corresponding author to find topic-expertise alignment.
90
91	1. Search PubMed for the PI's recent publications (last 5 years).
92	- Use `/search-lit` E-utilities:

People who liked this also viewed

6 picks

Type