How to Prioritize a Disease Indication

What Is an Indication Strategy?

Ask a portfolio or R&D strategy team to make the case for continuing (or dropping) work on an indication, and the real question is never "is this disease serious", it's whether the burden, the biology, the competition, and the deal-making all point the same direction. A biologically sound target with no patient segmentation is a weak case. A large market with no credible mechanism is a weak case. Indication strategy is what happens when all four get checked together instead of separately.

Patsnap Open calls the structured output of that check an Indication Strategy report, a ranked assessment with a scorecard attached, not just a database export. This guide walks through building one with Patsnap Open. No prior MCP experience required, by the end, you'll have connected three MCP servers and produced a scorecard you can hand to a portfolio review, with every claim traced back to a record.

What Data Do You Need for Indication Strategy?

The traditional version of this work is familiar to anyone who's built a portfolio case. Look up the disease definition and patient journey in a clinical reference. Search separately for epidemiology figures, checking that the population and year actually match. Read the target's biology in a database, then judge causality yourself. Check a trial registry for competition. Check a deal database for pricing signals. Each of those lives in a different kind of source, and none of them were built to talk to each other.

None of this is hard, exactly, it's just slow, and the slowness compounds. A high-burden disease with zero registered trials and zero matched deals isn't automatically a green light, it can just as easily mean the terminology doesn't match how the field is searched, or that commercial interest genuinely hasn't formed yet. The report has to say which, not just report a number.

How MCP Servers Automate the Scorecard

This is exactly the kind of work AI agents have gotten good at, reading across disease, epidemiology, target, trial and deal sources and holding every figure's population, year and geography straight, provided they can actually reach structured records instead of guessing at a number.

Patsnap Open exposes exactly this kind of tool over MCP, built on the same disease, epidemiology, target, trial and deal records used across Patsnap's life-science products. That matters here specifically: the agent isn't summarizing from memory, it's reading a structured record with a population, year and identifier attached, which is what lets the final scorecard cite its sources instead of asserting them.

How to Build an Indication Strategy Scorecard

Here's the whole mechanism in one picture: your AI client talks MCP to three Patsnap Open servers, each exposing one or more callable tools, and the tool results become the scorecard.

Fig. 2, One AI client, three MCP servers, five tools, one scorecard. Every arrow above is a real connection you'll make below.

Building it starts before the first tool call, connecting the three MCP servers. That takes five minutes:

1. Get a Patsnap Open API key. Sign up at the Patsnap Open Developer Portal, 10,000 free credits, no credit card required.
2. Open the MCP server pages. Visit Target & Disease, Clinical Trials, and Company & Deal Intelligence on the marketplace, and generate a connection URL from each using your API key.
3. Add all three URLs to an MCP-capable client. Paste them into your client's MCP/connector settings, Claude, ChatGPT, Cursor, or any client that supports remote MCP servers, and confirm the five tools below are visible.

Connect the following Patsnap Open MCP server(s) to this client:

Target & Disease: https://connect.patsnap.com/2a2645/logic-mcp?apikey=YOUR_API_KEY
Clinical Trials: https://connect.patsnap.com/051cd3/logic-mcp?apikey=YOUR_API_KEY
Company & Deal Intelligence: https://connect.patsnap.com/1f8934/logic-mcp?apikey=YOUR_API_KEY

Replace YOUR_API_KEY with my Patsnap Open API key. After connecting, confirm that disease_fetch, clinical_trial_search, and drug_deal_search are available. Do not begin the research workflow yet.

Once connected, five tools across those three servers cover the whole evidence chain:

Because the first three tools live on one server, one connection is enough to call all of them in the same conversation. From there, the calls run in a fixed order:

1. Resolve the indication with disease_fetch, Target & Disease, to get its definition, subtypes and patient journey.
2. Add burden evidence with epidemiology_search, same server, keeping population, geography and study year attached to every number.
3. Build the mechanistic case with target_fetch, same server, ending in a falsifiable hypothesis linking target, biomarker and expected effect.
4. Map the competitive field with clinical_trial_search on Clinical Trials, and check deal activity with drug_deal_search on Company & Deal Intelligence to see whether the space is already being priced.

Fill in the fields below, then paste the prompt into your connected AI client. The main request stays readable; the exact tool sequence is available as optional technical detail.

Using Patsnap Open's Target & Disease, Clinical Trials, and Company & Deal Intelligence connections, prepare an Indication Strategy for:

Indication: [INDICATION]
Geography: [GEOGRAPHY]
Target: [TARGET, IF KNOWN]
Decision: [PORTFOLIO / R&D / BUSINESS DEVELOPMENT]
Deal window: since [START DATE]

Assess the disease definition and patient journey, epidemiology and burden, unmet need, target rationale and safety, clinical competition, and relevant deal activity. Preserve population, geography, and study year for every epidemiology figure.

Return an executive assessment, disease background, burden evidence, unmet need, target rationale, clinical landscape, deal activity, strategic scorecard, evidence gaps, and next diligence priorities. Cite record identifiers, label directional counts, and state where evidence is insufficient.

Example: Indication Strategy Report

Two real scorecards show this end to end. Thymoma came back with 85 active or upcoming trial records but 0 KIT-targeted drug entries and 0 indication-specific deals since 2023, a biologically plausible target with essentially no commercial conviction yet. Presentey Anomaly came back with real epidemiology trend data and a 73/100 strategic score, against just 1 registered trial. Same five tools, two very different strategic pictures. Here's what the underlying report reads like when you open the richer one, an excerpt structured from the actual Presentey Anomaly evidence chain:

Fig. 3, The Indication Strategy dashboard built from this data: Presentey Anomaly’s strategic score breakdown and side-by-side scorecards comparing Presentey Anomaly and Thymoma.

The report itself doesn't have to stop at plain text. Once the evidence is in the conversation, you can ask the same AI client to turn it into a Word document, a slide deck, or a formatted HTML page, and you can feed in your own supplementary material (an internal portfolio note, a specific competitor watchlist) alongside the MCP results, so the output reflects both what Patsnap Open retrieved and what you already know.