|
Download README.md from Umesh1608/biointel: direct link, hf CLI and curl.
- Browser
- Download file 2.04 kB
-
https://huggingface.co/spaces/Umesh1608/biointel/resolve/main/README.md
- Command line
-
hf download hf://spaces/Umesh1608/biointel/README.md
-
curl -L -o README.md https://huggingface.co/spaces/Umesh1608/biointel/resolve/main/README.md
2.04 kB
| title: BioIntel | |
| emoji: 🧬 | |
| colorFrom: blue | |
| colorTo: green | |
| sdk: gradio | |
| sdk_version: 6.30.0 | |
| app_file: app.py | |
| pinned: false | |
| license: mit | |
| short_description: Biotech catalysts from SEC filings and trials | |
| # BioIntel | |
| Biotech catalyst intelligence built entirely from public sources. | |
| | Tab | What it shows | | |
| |---|---| | |
| | **Catalyst feed** | SEC 8-K filings matching 12 clinical/regulatory signal types — what already happened | | |
| | **Readout calendar** | Industry trials with a primary completion date ahead — dated catalysts that have not happened yet | | |
| | **Demand mix** | Modality mix of the upcoming pipeline, the closest public read on future wet-lab and manufacturing need | | |
| ## Sources | |
| - **SEC EDGAR full-text search** (`efts.sec.gov`) — no key, 10 req/s ceiling, descriptive User-Agent required | |
| - **ClinicalTrials.gov API v2** — no key | |
| Data is refreshed by a scheduled job and published as parquet to a Hugging Face | |
| Dataset repo, which this Space reads at startup. | |
| ## Method notes | |
| Catalyst phrases were chosen by **measurement rather than assumption**. Over a | |
| sample window, `"met its primary endpoint"` appeared in 54 filings while | |
| `"met the primary endpoint"` appeared in 19, and `"topline data"` outranked | |
| `"topline results"` — so each signal searches a group of variants. EDGAR | |
| accepts uppercase `OR` between quoted phrases; a `|` silently returns zero. | |
| Both APIs truncate silently: one EDGAR call returns at most 100 hits | |
| regardless of the reported total, and filtering trials client-side after a | |
| truncated fetch drops most of the result set. Both are paged and filtered | |
| server-side here. | |
| ## Limitations | |
| Signals co-occur — one filing can match both *endpoint met* and *endpoint | |
| missed*, which is a real reporting pattern, not a bug. Trial completion dates | |
| are plans and slip. Modality is inferred and reports `unknown` rather than | |
| guessing. CRO lab occupancy, private lease terms and options flow are **not** | |
| obtainable from public data. | |
| **An information tool built on public filings. Not investment advice.** | |