Initial dataset upload — Clinical Trials AI 2000-2025, generated by Gemma Miner

README.md

@@ -0,0 +1,517 @@
+---
+license: apache-2.0
+language:
+  - en
+size_categories:
+  - 1K<n<10K
+task_categories:
+  - tabular-classification
+  - tabular-regression
+  - text-classification
+tags:
+  - clinical-trials
+  - healthcare
+  - medical-ai
+  - artificial-intelligence
+  - machine-learning
+  - clinicaltrials-gov
+  - meta-research
+  - regulatory
+  - deployment
+pretty_name: Clinical Trials of AI / ML / Digital Health (2000 – 2025)
+configs:
+  - config_name: default
+    data_files:
+      - split: train
+        path: final_dataset.parquet
+---
+
+# Clinical Trials of AI / ML / Digital Health — 2000 → 2025
+
+> A typed, analysis-ready dataset of **3 000 clinical trials** registered on
+> **ClinicalTrials.gov** that involve artificial intelligence, machine
+> learning, or digital-health software, joined with **30 LLM-extracted
+> analytical variables** (use-type, disease area, sponsor mix, phase,
+> deployment score, evidence-strength score, responsible-AI keyword flags).
+>
+> Generated end-to-end (scrape → typed schema → per-row LLM extraction →
+> export) by **[Gemma Miner](https://github.com/moncifem/gemma-miner)** —
+> an autonomous text-to-dataset agent that turns any website into a
+> research-grade dataset in minutes.
+
+## TL;DR — what this dataset reveals
+
+- **AI-related clinical trials grew ~22× in 8 years.** From a steady < 30
+  / year through 2017, to **627 trials starting in 2025** alone. The
+  inflection is sharp and post-2017 — earlier than the LLM/ChatGPT wave.
+- **The clinical-AI ecosystem is now Chinese-led, not American.** China
+  hosts at least one site in **580 trials**, the United States in **500**.
+  Italy (201), France (187), Spain (147) and Türkiye (144) round out a
+  surprisingly even European top-5.
+- **63 % of trials are diagnostic, predictive, or imaging models** —
+  AI in healthcare is overwhelmingly about *deciding* (what is this? what
+  will happen?), not *treating* or *acting*.
+- **Only 1 % of trials reach Phase 3 or 4.** 91 % are coded "not
+  applicable" — these are mostly observational / device studies, not
+  drug-style efficacy RCTs. Translation: medical AI is being *measured*
+  much more than it is being *regulatorily approved*.
+- **The evidence pyramid is inverted.** Mean evidence-strength score is
+  **0.42** (pilot/feasibility-grade). Real-world deployment score sits at
+  **0.63** — many systems are *already in the clinic*, but only a minority
+  have pivotal-RCT evidence supporting them.
+- **Responsible-AI discourse is rare.** Only **1.6 %** of trial
+  descriptions mention "bias", **2.7 %** mention "privacy", **11.7 %**
+  mention "safety". Compare to **86 %** mentioning "AI" generally.
+- **Academic sponsors dominate (86 %)**, industry trails at 15 %. The top
+  sponsor — **Mayo Clinic, 42 trials** — has 3× more AI trials than any
+  for-profit. AI is being **studied by hospitals, sold by startups**.
+- **4 of the top 5 sponsors are Chinese institutions** (Sun Yat-sen
+  University, NTUH, Renmin Wuhan, CUHK), confirming the country-level
+  finding at the institutional level.
+
+## Quick start
+
+<details>
+<summary><b>📥 Load with 🤗 datasets</b> (click to expand)</summary>
+
+```python
+from datasets import load_dataset
+
+ds = load_dataset("moncefem/clinical-trials-ai-2000-2025", split="train")
+print(ds[0])
+print(ds.features)
+```
+</details>
+
+<details>
+<summary><b>🐼 Load with pandas (no `datasets` install needed)</b></summary>
+
+```python
+import pandas as pd
+
+df = pd.read_parquet(
+    "hf://datasets/moncefem/clinical-trials-ai-2000-2025/final_dataset.parquet"
+)
+print(df.shape)         # (3000, 30)
+print(df.dtypes)
+```
+</details>
+
+<details>
+<summary><b>🦆 Load with DuckDB (in-process SQL)</b></summary>
+
+```python
+import duckdb
+
+con = duckdb.connect()
+con.execute("""
+    CREATE VIEW trials AS
+    SELECT * FROM read_parquet(
+      'hf://datasets/moncefem/clinical-trials-ai-2000-2025/final_dataset.parquet'
+    )
+""")
+print(con.execute("""
+    SELECT ai_use_type, COUNT(*) AS n,
+           AVG(real_world_deployment_score) AS deployment,
+           AVG(evidence_strength_score)     AS evidence
+    FROM trials
+    WHERE ai_use_type IS NOT NULL
+    GROUP BY ai_use_type ORDER BY n DESC
+""").df())
+```
+</details>
+
+---
+
+## Charts at a glance
+
+### 1. AI in clinical trials is going vertical
+
+![Trials per start year](charts/trials_per_year.png)
+
+Until 2017, ClinicalTrials.gov logged fewer than 30 AI-related trials per
+year. Then: 60 (2018) → 89 (2019) → 216 (2020) → 284 (2021) → 344 (2022) →
+388 (2023) → **529 (2024) → 627 (2025)**.
+
+The inflection clearly **predates the generative-AI / LLM wave** — most of
+the growth is driven by classical ML, imaging models, and patient apps,
+not ChatGPT-era systems. (The `has_llm` flag is true on **only 1 of 3000
+trials** in the dataset.)
+
+<details>
+<summary><b>🔬 Reproduce this chart</b></summary>
+
+```python
+import pandas as pd, json, matplotlib.pyplot as plt
+df = pd.read_parquet("final_dataset.parquet")
+bronze = [json.loads(l) for l in open("dataset.jsonl", encoding="utf-8")]
+df = df.assign(start_year=[
+    int((r.get("start_date") or "")[:4]) if (r.get("start_date") or "")[:4].isdigit() else None
+    for r in bronze
+])
+yearly = df["start_year"].dropna().astype(int).value_counts().sort_index()
+yearly = yearly[(yearly.index >= 2005) & (yearly.index <= 2025)]
+yearly.plot.bar(figsize=(11, 5))
+plt.title("AI-related clinical trials — count by start year"); plt.show()
+```
+</details>
+
+### 2. AI in medicine is overwhelmingly about *deciding*
+
+![How AI is being used](charts/ai_use_type.png)
+
+| Use-type                  | Trials | Share |
+|---------------------------|--------|-------|
+| Diagnostic                | 870    | 31 %  |
+| Prediction / prognosis    | 612    | 21 %  |
+| Medical imaging           | 325    | 11 %  |
+| Treatment recommendation  | 325    | 11 %  |
+| Workflow                  | 156    | 5 %   |
+| Patient-facing app        | 116    | 4 %   |
+| Remote monitoring         | 42     | 1 %   |
+| Triage                    | 40     | 1 %   |
+| Wearable                  | 12     | 0 %   |
+
+Diagnostic + predictive + imaging models account for **63 %** of all
+trials. Patient-facing apps (apps used directly by patients), wearables
+and triage tools together are < 6 %. Read carefully: **clinical AI today
+is a tool for clinicians, not for patients**.
+
+### 3. The disease mix is concentrated
+
+![Therapeutic area distribution](charts/disease_area.png)
+
+Of trials with a clear primary disease area, **oncology dominates (22 %)**,
+followed by cardiology (13 %) and neurology (8 %). The "other / mixed"
+bucket holds 46 % — heterogeneous (rare diseases, infectious diseases,
+non-clinical decision support, multi-organ predictions).
+
+### 4. Geographic concentration is striking — and surprising
+
+![Top 15 countries hosting AI-related trials](charts/top_countries.png)
+
+**China (580 trials) has overtaken the United States (500)** as the top
+host country in the dataset, despite ClinicalTrials.gov being a US-run
+registry. Europe is highly distributed: Italy, France, Spain, Türkiye,
+the UK and Germany each host between 94 and 201 trials — none individually
+close to China or the US, but the European total is comparable.
+
+**96 distinct countries** are represented overall, though only **5.7 %**
+of trials are international (have sites in ≥ 2 countries) — the AI-trial
+ecosystem is mostly single-country.
+
+<details>
+<summary><b>🔬 Map your own geographic slice</b></summary>
+
+```python
+import pandas as pd, json
+from collections import Counter
+bronze = [json.loads(l) for l in open("dataset.jsonl", encoding="utf-8")]
+c = Counter()
+for r in bronze:
+    if isinstance(r.get("countries"), list):
+        for x in set(r["countries"]):
+            c[x] += 1
+print(pd.Series(dict(c)).sort_values(ascending=False).head(25))
+```
+</details>
+
+### 5. Sponsor mix: academic-led, all years
+
+![Sponsor mix per year](charts/sponsor_mix.png)
+
+Across every year of the dataset, **academic sponsors outnumber industry
+sponsors ~5-to-1**. Even at peak 2024-2025 volumes, industry-led trials
+remain a minority. This matters for translation: the studies producing
+evidence are *not* the same studies producing commercial products.
+
+**Top 5 sponsors** (by AI-related trial count):
+
+| Rank | Sponsor                              | Trials |
+|------|--------------------------------------|--------|
+| 1    | Mayo Clinic (US)                     | 42     |
+| 2    | Sun Yat-sen University (CN)          | 35     |
+| 3    | National Taiwan University Hospital  | 29     |
+| 4    | Renmin Hospital of Wuhan Univ. (CN)  | 26     |
+| 5    | Chinese University of Hong Kong      | 24     |
+
+**Four of the top five sponsors are based in China or Taiwan**, with
+only Mayo Clinic representing the US. This reinforces the country-level
+finding above — the AI-trial ecosystem visible on ClinicalTrials.gov is
+already Chinese-led at the institutional level too, not just by site
+count.
+
+(see `charts/top_sponsors.png` for the full top 15)
+
+![Top 15 sponsors](charts/top_sponsors.png)
+
+### 6. The maturity grid — most trials are "not applicable"
+
+![Phase × use-type heatmap](charts/maturity_grid.png)
+
+For every AI use-type, the modal trial phase is **"not applicable"** —
+because most medical-AI studies are observational or device-software
+studies that don't fit the drug-style Phase 1–4 framework. This is the
+single biggest reason to read deployment / evidence scores instead of
+relying on the phase field.
+
+**Only 1 % of trials reach Phase 3 or 4.** The interventional
+drug-style pipeline is not where AI gets evaluated.
+
+### 7. Deployment vs evidence — the inverted pyramid
+
+![Score distributions](charts/scores_distribution.png)
+
+Two LLM-derived scores tell different stories:
+
+- **Real-world deployment score** (left, mean **0.63**): a bimodal
+  distribution clustered around 0.5–0.8. Most trials are studying
+  systems that are **already used in clinical workflows** — not pure
+  research artifacts.
+- **Evidence-strength score** (right, mean **0.42**): peaks around 0.3
+  (pilot / feasibility) with a secondary lump at 0.6. Pivotal-RCT-grade
+  evidence (> 0.7) is rare.
+
+Combine those: **a lot of clinical AI is being deployed before it's been
+rigorously evaluated** — exactly the gap that motivates ongoing
+regulatory work (FDA SaMD, EU AI Act high-risk medical devices, MDR
+Class IIa/b).
+
+### 8. The "responsible AI" gap
+
+![Tech & responsibility mentions](charts/tech_mentions.png)
+
+Of the ten keyword flags extracted from trial descriptions:
+
+| Keyword          | Trials | Share |
+|------------------|--------|-------|
+| "AI"             | 2 578  | 86 %  |
+| "algorithm"      | 1 403  | 47 %  |
+| "software"       | 752    | 25 %  |
+| "machine learning"| 459   | 15 %  |
+| "deep learning"  | 385    | 13 %  |
+| "safety"         | 350    | 12 %  |
+| "mobile app"     | 330    | 11 %  |
+| "wearable"       | 168    | 6 %   |
+| "privacy"        | 80     | 3 %   |
+| "bias"           | 47     | **2 %** |
+
+Responsible-AI vocabulary is **two orders of magnitude rarer than
+"AI" itself** in trial descriptions. Whether this reflects authors not
+*writing* about bias/privacy (they may still test for it) or genuinely
+not *measuring* it is an empirical question this dataset is well-sized
+to study.
+
+---
+
+## Suggested research questions
+
+This dataset is sized for fast iteration on questions like:
+
+1. **Has the use-type distribution evolved over time?** Did diagnostic
+   models dominate forever, or did patient-apps / wearables rise (or
+   fall) in 2022-2025?
+2. **Does the deployment-vs-evidence gap differ by disease area?** Is
+   oncology AI more rigorously evaluated than cardiology AI?
+3. **Geographic specialisation:** does China focus more on imaging
+   trials? Does the US lead in patient-app trials?
+4. **Sponsor type vs evidence:** are industry-sponsored trials more
+   likely to be RCTs than academic ones — or less?
+5. **Where does responsible-AI vocabulary actually appear?** Is the
+   2 % "bias" share concentrated in a few disease areas (psychiatry,
+   dermatology) or evenly distributed?
+
+<details>
+<summary><b>🔬 Q1 sketch — use-type drift 2017 → 2025</b></summary>
+
+```python
+import pandas as pd, json
+df = pd.read_parquet("final_dataset.parquet")
+bronze = [json.loads(l) for l in open("dataset.jsonl", encoding="utf-8")]
+df["start_year"] = [
+    int((r.get("start_date") or "")[:4]) if (r.get("start_date") or "")[:4].isdigit() else None
+    for r in bronze
+]
+df["era"] = df["start_year"].apply(
+    lambda y: "≤2017" if (y or 0) <= 2017
+    else ("2018-2021" if (y or 0) <= 2021 else "2022-2025"))
+ct = (df.groupby(["era", "ai_use_type"]).size()
+        .unstack(fill_value=0)
+        .apply(lambda r: (r / r.sum() * 100).round(1), axis=1))
+print(ct.to_string())
+```
+</details>
+
+<details>
+<summary><b>🔬 Q4 sketch — industry vs academic, evidence quality</b></summary>
+
+```python
+import pandas as pd
+df = pd.read_parquet("final_dataset.parquet")
+df["sponsor_class"] = (
+    df["industry_sponsored"].fillna(False).astype(int)*2 +
+    df["academic_sponsored"].fillna(False).astype(int)
+).map({3: "both", 2: "industry only", 1: "academic only", 0: "neither/gov"})
+print(df.groupby("sponsor_class").agg(
+    n=("id", "count"),
+    rct_pct=("has_randomization", lambda s: (s == True).mean() * 100),
+    evidence=("evidence_strength_score", "mean"),
+    deployment=("real_world_deployment_score", "mean"),
+).round(2).to_string())
+```
+</details>
+
+---
+
+## Codebook (30 silver columns + 20 bronze metadata columns)
+
+### Silver — LLM-extracted analytical variables
+
+| Column                          | Type     | Description |
+|---------------------------------|----------|-------------|
+| `id`                            | string   | Deterministic content-hash id |
+| `ai_use_type`                   | enum     | diagnosis · prediction · imaging · treatment_recommendation · workflow · patient_app · remote_monitoring · triage · wearable · other |
+| `disease_area`                  | enum     | oncology · cardiology · neurology · endocrinology · radiology · other |
+| `trial_phase_bucket`            | enum     | not_applicable · early_phase · phase_1 · phase_2 · phase_3 · phase_4 |
+| `is_completed`                  | boolean  | Trial status = completed |
+| `is_recruiting`                 | boolean  | Trial status = recruiting |
+| `industry_sponsored`            | boolean  | Has at least one industry sponsor / collaborator |
+| `academic_sponsored`            | boolean  | Has at least one academic / hospital sponsor |
+| `government_sponsored`          | boolean  | Has at least one government / NIH-like funder |
+| `has_randomization`             | boolean  | Description indicates randomisation |
+| `has_blinding`                  | boolean  | Description indicates single / double / triple blinding |
+| `enrollment_bucket`             | enum     | small · medium · large · very_large |
+| `includes_children`             | boolean  | Eligibility includes minors |
+| `includes_older_adults`         | boolean  | Eligibility includes 65 + |
+| `sex_all`                       | boolean  | Eligibility = ALL sexes |
+| `country_count`                 | integer  | # distinct countries hosting sites |
+| `international_trial`           | boolean  | `country_count` ≥ 2 |
+| `mentions_algorithm`            | boolean  | Description text contains "algorithm" |
+| `mentions_machine_learning`     | boolean  | … "machine learning" |
+| `mentions_deep_learning`        | boolean  | … "deep learning" |
+| `mentions_ai`                   | boolean  | … "AI" / "artificial intelligence" |
+| `mentions_software`             | boolean  | … "software" |
+| `mentions_mobile_app`           | boolean  | … "mobile app" |
+| `mentions_wearable`             | boolean  | … "wearable" |
+| `mentions_bias`                 | boolean  | … "bias" |
+| `mentions_safety`               | boolean  | … "safety" |
+| `mentions_privacy`              | boolean  | … "privacy" |
+| `real_world_deployment_score`   | float    | 0–1 score: how close to clinic deployment (LLM judgement on metadata + summary) |
+| `evidence_strength_score`       | float    | 0–1 score: how rigorous the planned evidence is (pilot ≈ 0.2 → pivotal RCT ≈ 0.9) |
+| `has_llm`                       | boolean  | Description specifically mentions LLM / GPT / Claude / Gemini |
+
+### Bronze — original ClinicalTrials.gov metadata (joined by row position)
+
+`trial_url`, `nct_id`, `title`, `conditions`, `interventions`, `sponsor`,
+`collaborators`, `study_type`, `phase`, `enrollment`, `start_date`,
+`completion_date`, `status`, `countries`, `locations`,
+`eligibility_criteria`, `age_range`, `sex`, `outcomes`, `brief_summary`.
+
+(Both layers ship together — the parquet contains the silver; the bronze
+JSONL is in `dataset.jsonl` for the in-repo source.)
+
+---
+
+## How this dataset was built
+
+This file was produced by **[Gemma Miner](https://github.com/moncifem/gemma-miner)**
+in a single autonomous agent run:
+
+1. **Harvest** — agent paginated ClinicalTrials.gov's listing API,
+   pulling 3 000 AI/ML/digital-health trials with their full metadata.
+2. **Codebook design** — an LLM proposed 30 typed variables matching the
+   analytical brief (use-type taxonomy, disease area, sponsor flags,
+   maturity/evidence scores, responsible-AI keyword detectors).
+3. **Per-row extraction** — for each trial, an LLM read the title +
+   conditions + interventions + brief_summary and emitted a JSON object
+   conforming to the codebook; the system then deterministically
+   coerced values (booleans normalised, ambiguous → null, enums snapped
+   to nearest valid value).
+4. **Export** — parquet + CSV + this card + charts.
+
+No fine-tuning. No labelled training data. Reproducible.
+
+<details>
+<summary><b>🔬 Rebuild this dataset from scratch</b></summary>
+
+```bash
+pip install gemma-miner          # from https://github.com/moncifem/gemma-miner
+export OPENROUTER_API_KEY=...    # any OpenAI-compatible provider
+gemma42                          # drops into the REPL
+
+# in the REPL:
+> Build me a statistics-ready dataset of 3000 AI / ML / digital-health
+  clinical trials from ClinicalTrials.gov, with sponsor mix, disease
+  area, trial-phase bucket, AI use-type taxonomy, real-world deployment
+  score and evidence-strength score.
+```
+</details>
+
+---
+
+## Limitations & honest caveats
+
+- **LLM-derived columns** were extracted from each trial's
+  ClinicalTrials.gov metadata + `brief_summary` (≤ 2 KB of English text).
+  A `true` flag is high-precision; a `null` / `false` means the summary
+  didn't mention the concept — **not** that the trial doesn't address it.
+- **Scores are heuristics.** `real_world_deployment_score` and
+  `evidence_strength_score` are calibrated against the LLM's prior on
+  what a "deployed" or "rigorous" trial looks like — they're useful for
+  *ranking* trials within the dataset, not as absolute ground-truth.
+- **Sample = AI-related trials**, not all clinical trials. The sample
+  was selected by the underlying ClinicalTrials.gov search; comparisons
+  to non-AI trials require a separate dataset.
+- **Date parsing**: ClinicalTrials.gov mixes `YYYY-MM-DD`, `YYYY-MM`
+  and `YYYY` strings. The `start_year` field is robust but you should
+  not assume month-level precision.
+- **Bronze ↔ silver join is by ROW POSITION**, not id — the silver
+  table's `id` is a content hash and does not appear in bronze. The two
+  files are aligned 1-to-1 (both 3000 rows).
+- **No de-duplication across protocol amendments.** A few NCT IDs may
+  appear with multiple versions; we keep the first occurrence.
+- **The `phase` field is messy.** ClinicalTrials.gov's "NA"/"N/A" labels
+  are common for device studies; this is why `trial_phase_bucket`
+  collapses them into `not_applicable` rather than a numeric phase.
+
+---
+
+## Citation
+
+```bibtex
+@misc{elmouden_clinical_trials_ai_2025,
+  title  = {Clinical Trials of AI / ML / Digital Health (2000-2025)},
+  author = {EL-Mouden, Moncif},
+  year   = {2025},
+  note   = {Generated by Gemma Miner from https://clinicaltrials.gov},
+  url    = {https://huggingface.co/datasets/moncefem/clinical-trials-ai-2000-2025},
+}
+
+@software{elmouden_gemma_miner_2025,
+  title  = {Gemma Miner: an autonomous text-to-dataset agent},
+  author = {EL-Mouden, Moncif and contributors},
+  year   = {2025},
+  url    = {https://github.com/moncifem/gemma-miner},
+}
+```
+
+Underlying trial records are published by the U.S. National Library of
+Medicine on <https://clinicaltrials.gov>; consult those records for the
+authoritative protocols and outcomes.
+
+## Author & links
+
+- 👤 **Moncif EL-Mouden** — [🤗 huggingface.co/moncefem](https://huggingface.co/moncefem)
+- 🤖 **Gemma Miner** (the generator) — <https://github.com/moncifem/gemma-miner>
+- 🇺🇸 **Source** — <https://clinicaltrials.gov>
+
+## License
+
+[**Apache License 2.0**](https://www.apache.org/licenses/LICENSE-2.0).
+
+Please attribute:
+
+- **ClinicalTrials.gov** (U.S. National Library of Medicine) as the
+  source of the underlying trial records, and
+- **Gemma Miner** (<https://github.com/moncifem/gemma-miner>) as the
+  dataset generator.

_make_charts.py

@@ -0,0 +1,312 @@
+"""Deep analysis of the Clinical Trials AI dataset.
+
+Joins bronze (ClinicalTrials.gov metadata) with silver (LLM-extracted typed
+variables) by row position and produces 9 charts + insights.json.
+"""
+from __future__ import annotations
+
+import json
+from collections import Counter
+from pathlib import Path
+
+import matplotlib
+
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+import numpy as np
+import pandas as pd
+
+HERE = Path(__file__).parent
+CHARTS = HERE / "charts"
+CHARTS.mkdir(exist_ok=True)
+
+plt.rcParams.update({
+    "figure.dpi": 130, "savefig.dpi": 130, "savefig.bbox": "tight",
+    "font.family": "DejaVu Sans",
+    "axes.spines.top": False, "axes.spines.right": False,
+    "axes.grid": True, "axes.grid.axis": "y",
+    "grid.color": "#e5e7eb", "grid.linestyle": "-", "grid.linewidth": 0.8,
+    "axes.titlesize": 14, "axes.titleweight": "bold", "axes.labelsize": 11,
+})
+
+ACCENT = "#2563eb"
+ACCENT_2 = "#dc2626"
+ACCENT_3 = "#16a34a"
+ACCENT_4 = "#f59e0b"
+NEUTRAL = "#6b7280"
+
+# ── Load + JOIN ────────────────────────────────────────────────────────────
+silver = pd.read_parquet(HERE / "final_dataset.parquet")
+bronze_path = HERE.parent / "dataset.jsonl"
+bronze_rows = [json.loads(l) for l in bronze_path.open("r", encoding="utf-8")]
+bronze = pd.DataFrame(bronze_rows)
+print(f"silver: {len(silver)} × {len(silver.columns)}")
+print(f"bronze: {len(bronze)} × {len(bronze.columns)}")
+assert len(silver) == len(bronze)
+df = pd.concat([bronze.reset_index(drop=True), silver.reset_index(drop=True)], axis=1)
+df = df.loc[:, ~df.columns.duplicated()]
+print(f"joined: {len(df)} × {len(df.columns)}")
+
+
+def _year(s):
+    if not isinstance(s, str):
+        return None
+    s = s.strip()
+    if len(s) >= 4 and s[:4].isdigit():
+        try:
+            y = int(s[:4])
+            if 1990 <= y <= 2030:
+                return y
+        except ValueError:
+            pass
+    return None
+
+
+df["start_year"] = df["start_date"].apply(_year)
+
+# Friendly maps
+USE_MAP = {
+    "diagnosis": "Diagnostic",
+    "prediction": "Prediction / prognosis",
+    "imaging": "Medical imaging",
+    "treatment_recommendation": "Treatment recommendation",
+    "workflow": "Clinical workflow",
+    "patient_app": "Patient-facing app",
+    "remote_monitoring": "Remote monitoring",
+    "triage": "Triage",
+    "wearable": "Wearable",
+    "other": "Other",
+}
+DISEASE_MAP = {
+    "oncology": "Oncology",
+    "cardiology": "Cardiology",
+    "neurology": "Neurology",
+    "endocrinology": "Endocrinology",
+    "radiology": "Radiology",
+    "other": "Other / mixed",
+}
+
+# ── 1. Trials per start year ───────────────────────────────────────────────
+yearly = df["start_year"].dropna().astype(int).value_counts().sort_index()
+yearly = yearly[(yearly.index >= 2005) & (yearly.index <= 2025)]
+fig, ax = plt.subplots(figsize=(11, 5))
+ax.bar(yearly.index, yearly.values, color=ACCENT)
+ax.set_title("AI-related clinical trials — count by start year")
+ax.set_xlabel("Start year"); ax.set_ylabel("Trials started")
+ax.set_xticks(yearly.index); ax.tick_params(axis="x", rotation=45)
+peak_yr, peak_n = yearly.idxmax(), yearly.max()
+ax.annotate(f"peak: {peak_yr}, {peak_n} trials",
+            xy=(peak_yr, peak_n), xytext=(peak_yr - 4, peak_n - 30),
+            arrowprops=dict(arrowstyle="->", color=NEUTRAL),
+            fontsize=10, color=NEUTRAL)
+fig.savefig(CHARTS / "trials_per_year.png"); plt.close(fig)
+print("  ✓ trials_per_year.png")
+
+# ── 2. AI use-type taxonomy ───────────────────────────────────────────────
+use = df["ai_use_type"].dropna().value_counts()
+use = use.rename(index=lambda x: USE_MAP.get(x, x))
+fig, ax = plt.subplots(figsize=(9, 5))
+bars = ax.barh(use.index[::-1], use.values[::-1], color=ACCENT)
+ax.set_title("How AI is being used in clinical trials")
+ax.set_xlabel(f"Number of trials (n={int(use.sum())})")
+ax.grid(axis="y", visible=False); ax.grid(axis="x", visible=True)
+for bar, val in zip(bars, use.values[::-1]):
+    ax.text(val + 8, bar.get_y() + bar.get_height() / 2,
+            f"{val}  ({val / use.sum():.0%})",
+            va="center", fontsize=9, color=NEUTRAL)
+fig.savefig(CHARTS / "ai_use_type.png"); plt.close(fig)
+print("  ✓ ai_use_type.png")
+
+# ── 3. Disease area ────────────────────────────────────────────────────────
+disease = df["disease_area"].dropna().value_counts()
+disease = disease.rename(index=lambda x: DISEASE_MAP.get(x, x))
+fig, ax = plt.subplots(figsize=(9, 4.5))
+bars = ax.barh(disease.index[::-1], disease.values[::-1], color=ACCENT_3)
+ax.set_title("Therapeutic area distribution")
+ax.set_xlabel(f"Number of trials (n={int(disease.sum())})")
+ax.grid(axis="y", visible=False); ax.grid(axis="x", visible=True)
+for bar, val in zip(bars, disease.values[::-1]):
+    ax.text(val + 8, bar.get_y() + bar.get_height() / 2,
+            f"{val}  ({val / disease.sum():.0%})",
+            va="center", fontsize=9, color=NEUTRAL)
+fig.savefig(CHARTS / "disease_area.png"); plt.close(fig)
+print("  ✓ disease_area.png")
+
+# ── 4. Sponsor mix over time ──────────────────────────────────────────────
+years = sorted(int(y) for y in df["start_year"].dropna().unique() if 2010 <= y <= 2024)
+ind, aca, gov = [], [], []
+for y in years:
+    sub = df[df["start_year"] == y]
+    ind.append(int((sub["industry_sponsored"] == True).sum()))
+    aca.append(int((sub["academic_sponsored"] == True).sum()))
+    gov.append(int((sub["government_sponsored"] == True).sum()))
+fig, ax = plt.subplots(figsize=(11, 5))
+ax.bar(years, aca, label="academic", color=ACCENT)
+ax.bar(years, ind, bottom=aca, label="industry", color=ACCENT_2)
+ax.bar(years, gov, bottom=[a + i for a, i in zip(aca, ind)],
+       label="government", color=ACCENT_4)
+ax.set_title("Sponsor mix per year (trials may have multiple sponsor types)")
+ax.set_xlabel("Start year"); ax.set_ylabel("Trials")
+ax.set_xticks(years); ax.tick_params(axis="x", rotation=45)
+ax.legend(frameon=False, loc="upper left")
+fig.savefig(CHARTS / "sponsor_mix.png"); plt.close(fig)
+print("  ✓ sponsor_mix.png")
+
+# ── 5. Phase × AI use heatmap ──────────────────────────────────────────────
+phase_order = ["not_applicable", "early_phase", "phase_1", "phase_2", "phase_3", "phase_4"]
+use_top = list(df["ai_use_type"].value_counts().head(7).index)
+pt = (
+    df[df["ai_use_type"].isin(use_top) & df["trial_phase_bucket"].isin(phase_order)]
+    .groupby(["ai_use_type", "trial_phase_bucket"]).size()
+    .unstack(fill_value=0)
+    .reindex(index=use_top, columns=phase_order, fill_value=0)
+)
+fig, ax = plt.subplots(figsize=(9, 5))
+im = ax.imshow(pt.values, aspect="auto", cmap="Blues")
+ax.set_xticks(range(len(phase_order)))
+ax.set_xticklabels([p.replace("_", " ") for p in phase_order], rotation=30, ha="right")
+ax.set_yticks(range(len(use_top)))
+ax.set_yticklabels([USE_MAP.get(u, u) for u in use_top])
+ax.set_title("Trial phase × AI use-type (counts)")
+for i in range(len(use_top)):
+    for j in range(len(phase_order)):
+        v = pt.values[i, j]
+        if v > 0:
+            ax.text(j, i, int(v), ha="center", va="center",
+                    color="white" if v > pt.values.max() * 0.5 else "black",
+                    fontsize=9)
+ax.grid(False)
+fig.colorbar(im, ax=ax, label="trials")
+fig.savefig(CHARTS / "maturity_grid.png"); plt.close(fig)
+print("  ✓ maturity_grid.png")
+
+# ── 6. Score distributions ─────────────────────────────────────────────────
+ds = df["real_world_deployment_score"].dropna()
+es = df["evidence_strength_score"].dropna()
+fig, axes = plt.subplots(1, 2, figsize=(11, 4.5))
+axes[0].hist(ds, bins=20, color=ACCENT, edgecolor="white")
+axes[0].set_title("Real-world deployment score")
+axes[0].set_xlabel("0 = research artifact · 1 = deployed in clinic")
+axes[0].set_ylabel("Trials")
+axes[0].axvline(ds.mean(), color=ACCENT_2, linewidth=2, label=f"mean {ds.mean():.2f}")
+axes[0].axvline(ds.median(), color="black", linewidth=1.5, linestyle="--",
+                label=f"median {ds.median():.2f}")
+axes[0].legend(frameon=False)
+axes[1].hist(es, bins=20, color=ACCENT_3, edgecolor="white")
+axes[1].set_title("Evidence-strength score")
+axes[1].set_xlabel("0 = pilot/feasibility · 1 = pivotal RCT")
+axes[1].set_ylabel("Trials")
+axes[1].axvline(es.mean(), color=ACCENT_2, linewidth=2, label=f"mean {es.mean():.2f}")
+axes[1].axvline(es.median(), color="black", linewidth=1.5, linestyle="--",
+                label=f"median {es.median():.2f}")
+axes[1].legend(frameon=False)
+fig.savefig(CHARTS / "scores_distribution.png"); plt.close(fig)
+print("  ✓ scores_distribution.png")
+
+# ── 7. Tech mentions ──────────────────────────────────────────────────────
+mention_cols = [
+    "mentions_ai", "mentions_algorithm", "mentions_machine_learning",
+    "mentions_deep_learning", "mentions_software", "mentions_mobile_app",
+    "mentions_wearable", "mentions_safety", "mentions_bias", "mentions_privacy",
+]
+m_counts = {c: int((df[c] == True).sum()) for c in mention_cols}
+m_counts = dict(sorted(m_counts.items(), key=lambda kv: kv[1]))
+pretty_m = {
+    "mentions_ai": '"AI"',
+    "mentions_algorithm": '"algorithm"',
+    "mentions_machine_learning": '"machine learning"',
+    "mentions_deep_learning": '"deep learning"',
+    "mentions_software": '"software"',
+    "mentions_mobile_app": '"mobile app"',
+    "mentions_wearable": '"wearable"',
+    "mentions_safety": '"safety"',
+    "mentions_bias": '"bias"',
+    "mentions_privacy": '"privacy"',
+}
+labels = [pretty_m.get(k, k) for k in m_counts]
+vals = list(m_counts.values())
+fig, ax = plt.subplots(figsize=(9, 5))
+bars = ax.barh(labels, vals, color=ACCENT)
+ax.set_title("Tech & responsibility keywords mentioned in trial descriptions")
+ax.set_xlabel(f"# trials mentioning the term (n={len(df)})")
+ax.grid(axis="y", visible=False); ax.grid(axis="x", visible=True)
+for bar, v in zip(bars, vals):
+    ax.text(v + 20, bar.get_y() + bar.get_height() / 2,
+            f"{v}  ({v / len(df):.0%})",
+            va="center", fontsize=9, color=NEUTRAL)
+fig.savefig(CHARTS / "tech_mentions.png"); plt.close(fig)
+print("  ✓ tech_mentions.png")
+
+# ── 8. Top sponsors ────────────────────────────────────────────────────────
+sponsors = df["sponsor"].dropna().value_counts().head(15)
+fig, ax = plt.subplots(figsize=(10, 5.5))
+bars = ax.barh(sponsors.index[::-1], sponsors.values[::-1], color=ACCENT)
+ax.set_title("Top 15 sponsors by AI-related trial count")
+ax.set_xlabel("Trials")
+ax.grid(axis="y", visible=False); ax.grid(axis="x", visible=True)
+for bar, val in zip(bars, sponsors.values[::-1]):
+    ax.text(val + 1, bar.get_y() + bar.get_height() / 2, str(int(val)),
+            va="center", fontsize=9, color=NEUTRAL)
+fig.savefig(CHARTS / "top_sponsors.png"); plt.close(fig)
+print("  ✓ top_sponsors.png")
+
+# ── 9. Top countries ──────────────────────────────────────────────────────
+country_counter: Counter = Counter()
+for cs in df["countries"].dropna():
+    if isinstance(cs, list):
+        seen = set()
+        for c in cs:
+            if isinstance(c, str) and c not in seen:
+                country_counter[c] += 1
+                seen.add(c)
+top_c = country_counter.most_common(15)
+fig, ax = plt.subplots(figsize=(10, 5.5))
+labels = [c for c, _ in top_c][::-1]
+vals = [n for _, n in top_c][::-1]
+bars = ax.barh(labels, vals, color=ACCENT_3)
+ax.set_title("Top 15 countries hosting AI-related trials")
+ax.set_xlabel("Trials with at least one site in this country")
+ax.grid(axis="y", visible=False); ax.grid(axis="x", visible=True)
+for bar, val in zip(bars, vals):
+    ax.text(val + 5, bar.get_y() + bar.get_height() / 2, str(int(val)),
+            va="center", fontsize=9, color=NEUTRAL)
+fig.savefig(CHARTS / "top_countries.png"); plt.close(fig)
+print("  ✓ top_countries.png")
+
+# ── insights ───────────────────────────────────────────────────────────────
+insights = {
+    "n_trials": int(len(df)),
+    "year_range": [int(df["start_year"].dropna().astype(int).min()),
+                   int(df["start_year"].dropna().astype(int).max())],
+    "year_with_most": [int(peak_yr), int(peak_n)],
+    "trials_2020_2024": int(((df["start_year"] >= 2020) & (df["start_year"] <= 2024)).sum()),
+    "share_diagnostic_or_prediction_pct": round(
+        100 * (df["ai_use_type"].isin(["diagnosis", "prediction", "imaging"])).sum() / len(df), 1),
+    "share_phase_3_4_pct": round(
+        100 * df["trial_phase_bucket"].isin(["phase_3", "phase_4"]).sum() / len(df), 1),
+    "share_not_applicable_phase_pct": round(
+        100 * (df["trial_phase_bucket"] == "not_applicable").sum() / len(df), 1),
+    "share_with_randomization_pct": round(
+        100 * (df["has_randomization"] == True).sum() / len(df), 1),
+    "share_industry_sponsored_pct": round(
+        100 * (df["industry_sponsored"] == True).sum() / len(df), 1),
+    "share_academic_sponsored_pct": round(
+        100 * (df["academic_sponsored"] == True).sum() / len(df), 1),
+    "share_international_pct": round(
+        100 * (df["country_count"] >= 2).sum() / len(df), 1),
+    "mean_deployment_score": round(float(ds.mean()), 3),
+    "mean_evidence_score": round(float(es.mean()), 3),
+    "share_mentions_bias_pct": round(100 * m_counts["mentions_bias"] / len(df), 1),
+    "share_mentions_safety_pct": round(100 * m_counts["mentions_safety"] / len(df), 1),
+    "share_mentions_privacy_pct": round(100 * m_counts["mentions_privacy"] / len(df), 1),
+    "top_oncology_pct": round(100 * (df["disease_area"] == "oncology").sum() / len(df), 1),
+    "top_sponsor": str(sponsors.index[0]),
+    "top_sponsor_count": int(sponsors.values[0]),
+    "top_country": top_c[0][0],
+    "top_country_count": int(top_c[0][1]),
+    "n_countries_total": int(len(country_counter)),
+}
+(HERE / "insights.json").write_text(
+    json.dumps(insights, indent=2, ensure_ascii=False, default=str)
+)
+print("\ninsights:", json.dumps(insights, indent=2, default=str))

codebook.md

@@ -0,0 +1,31 @@
+# Codebook: Clinical Trials AI
+
+- **id**: Extracted variable
+- **ai_use_type**: Extracted variable
+- **disease_area**: Extracted variable
+- **trial_phase_bucket**: Extracted variable
+- **is_completed**: Extracted variable
+- **is_recruiting**: Extracted variable
+- **industry_sponsored**: Extracted variable
+- **academic_sponsored**: Extracted variable
+- **government_sponsored**: Extracted variable
+- **has_randomization**: Extracted variable
+- **has_blinding**: Extracted variable
+- **enrollment_bucket**: Extracted variable
+- **includes_children**: Extracted variable
+- **includes_older_adults**: Extracted variable
+- **sex_all**: Extracted variable
+- **country_count**: Extracted variable
+- **international_trial**: Extracted variable
+- **mentions_algorithm**: Extracted variable
+- **mentions_machine_learning**: Extracted variable
+- **mentions_deep_learning**: Extracted variable
+- **mentions_ai**: Extracted variable
+- **mentions_software**: Extracted variable
+- **mentions_mobile_app**: Extracted variable
+- **mentions_wearable**: Extracted variable
+- **mentions_bias**: Extracted variable
+- **mentions_safety**: Extracted variable
+- **mentions_privacy**: Extracted variable
+- **real_world_deployment_score**: Extracted variable
+- **evidence_strength_score**: Extracted variable

codebook.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:3d509ae87deb2c14fc3fdb05778c685a6222cd6c87144a557d85121341b98b0d
+size 3726300

final_dataset.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2c0be8eb0adb3ae234a8e0acca1f2d50b0fa50bb0e8e05faea3720758c08da64
+size 84627

insights.json

@@ -0,0 +1,30 @@
+{
+  "n_trials": 3000,
+  "year_range": [
+    2000,
+    2028
+  ],
+  "year_with_most": [
+    2025,
+    627
+  ],
+  "trials_2020_2024": 1761,
+  "share_diagnostic_or_prediction_pct": 60.2,
+  "share_phase_3_4_pct": 1.0,
+  "share_not_applicable_phase_pct": 90.6,
+  "share_with_randomization_pct": 22.2,
+  "share_industry_sponsored_pct": 15.0,
+  "share_academic_sponsored_pct": 85.7,
+  "share_international_pct": 5.7,
+  "mean_deployment_score": 0.632,
+  "mean_evidence_score": 0.415,
+  "share_mentions_bias_pct": 1.6,
+  "share_mentions_safety_pct": 11.7,
+  "share_mentions_privacy_pct": 2.7,
+  "top_oncology_pct": 22.2,
+  "top_sponsor": "Mayo Clinic",
+  "top_sponsor_count": 42,
+  "top_country": "China",
+  "top_country_count": 580,
+  "n_countries_total": 96
+}

report.md

@@ -0,0 +1,3 @@
+# How much medical AI is actually being clinically tested?
+
+This dataset contains 3000 clinical trials related to AI, machine learning, and digital health. Analysis shows a significant focus on diagnostic algorithms and patient-facing applications.