Custom Datasets
Customer Stories

A custom dataset, built to spec and updated weekly

Sleuth builds a dataset to your spec, checks it, and delivers it weekly. This one helps a health-tech company show hospitals what they need to prepare for new, advanced therapies.

A health-tech company

A custom dataset, built to spec and updated weekly

Download the full case study

Partner

Hospital administrators use its platform to choose which advanced therapies to offer and when to begin months-long preparation for them.

Challenge

The platform needs current data on upcoming therapies and what hospitals need to offer them. Public sources lack the custom fields and use inconsistent formats. The team had no analysts or engineers to build the dataset.

Approach

Outcome

The columns your decisions need, even when no database carries them

Their spec covers late-stage cell and gene therapies facing an FDA decision within 24 months, with sponsor, phase, FDA status, dates, and trials. Six custom columns add what hospitals need to prepare.

Six custom columns: hospital preparation lead time, infrastructure tier, eligible patients per academic center, expected price tier, reimbursement clarity, and approval confidence, each with an illustrative value.
Values shown are illustrative.

Each value is built from many sources

No single source holds values like these. Sleuth’s research agents combine evidence from relevant documents, including paid sources, into each value. The dataset includes the reasoning and sources behind it.

Labels, trial protocols, sponsor news, conference talks, and broker research flowing through Sleuth’s research agents into one value.

The reasoning and sources behind each value

One opened value in the dataset, with its reasoning and the sources behind it.
Each custom field includes this detail in every release. This example is illustrative.

Checks built to your spec, run on every release

Every Wednesday, a new release lands in their platform. Before it ships, it runs through checks Sleuth wrote from their spec at the start, on top of the knowledge graph’s own. Four of the problems they catch:

Four checks run on every release: matching duplicate names, resolving sources that disagree, re-researching a passed decision date, and updating related columns.

Add fields as needs change

When hospitals asked which nearby centers ran each trial, Sleuth added trial-site data to the next release. Brand names followed.

Their hospitals see what’s coming every week, and no one on their team builds or maintains the data.

Download
Use case

Tell us the dataset your team needs

Bring your own asset, target, or indication, and Sleuth builds the analysis for you.

More from Sleuth