Patent vulnerability score — methodology
Score bands
| Band | Score | Interpretation |
|---|---|---|
| Ironclad | 0-24 | Composition-of-matter, broad genus claims, recent issuance with long runway. Successful IPR challenge unlikely; design-around requires structurally different molecule. |
| Strong | 25-49 | Substantive scope (method of treatment, formulation with non-obvious limitation) with reasonable runway. IPR feasible but expensive; design-around often requires reformulation. |
| Moderate | 50-69 | Method of use, polymorph, or formulation patent with prior-art exposure or narrow claim scope. IPRs in this band invalidate or narrow at industry baseline rates. |
| Vulnerable | 70-100 | Method of use with thin novelty, late-issued continuation, narrow polymorph, salt form, or process patent. High-value targets for paragraph IV first-filers and biosimilar developers. |
The formula
The score blends four factors. Each contributes a sub-score that's combined linearly and clamped to 0-100.
1. Claim type (50% weight)
The base vulnerability bias for each FDA Orange Book classification + USPTO claim type. These are calibrated against historical PTAB IPR outcomes (2012-2025) where the patent class was the controlled variable.
| Classification | Base score | Why |
|---|---|---|
| Composition of matter (drug substance) | 15 | Hardest to invalidate. Novelty + non-obviousness directly tied to molecular structure. |
| Genus / broad chemical claim | 20 | Strong but expandable target if prior-art genus is found. |
| Formulation (drug product, novel excipient) | 45 | Vulnerable to design-around with alternative excipients; obvious-to-try defences common. |
| Polymorph / crystalline form | 55 | Often successfully challenged where prior art shows the form was inherent or trivially produced. |
| Salt form | 60 | Frequently invalidated under In re Adamantyl / obvious salts doctrine. |
| Method of use (indication) | 65 | Section viii carve-outs + skinny-labeling provide design-arounds. High IPR success rate. |
| Process / synthesis | 70 | Different synthesis route generally suffices. |
| Combination / dose regimen | 72 | Often vulnerable to obviousness given known mono-therapy + known co-therapy. |
2. Age (U-curve, 20% weight)
Patents at the extremes of their lifecycle are more vulnerable: very recent grants can still be challenged for prior art that didn't surface at examination, and patents near expiry have less remaining commercial value (so generic filers can wait it out cheaply, but if they file IPR the panel knows the patent is on the way out).
- Issued < 2 yrs ago: +6 (post-grant review window open, fresh challenges)
- Issued 2-5 yrs ago: +2
- Issued 5-10 yrs ago: 0 (baseline — middle-aged patents have settled prior-art landscape)
- Issued 10-15 yrs ago: +3
- Issued > 15 yrs ago: +6 (often continuation chains with thin claims)
3. Family size (15% weight)
Counter-intuitively, larger international families correlate with higher vulnerability. Large families typically indicate continuation/divisional strategy aimed at evergreening — narrower secondary claims layered onto a primary patent. Those secondary claims are statistically more attackable.
- 1-3 family members: -3 (focused, often comp-of-matter)
- 4-10 family members: 0
- 11-30 family members: +4
- > 30 family members: +8 (evergreening signature)
4. Exclusivity runway (15% weight)
Time remaining on the patent. Patents with little time left have less ROI on a paragraph IV / IPR challenge, but also tend to be the secondary patents added late in the lifecycle. Patents with very long runways protect very early-lifecycle drugs where strategic challenges are uncommon.
- < 2 yrs to expiry: +6 (challenge ROI low, but these are often the thinnest secondary patents)
- 2-5 yrs to expiry: +4 (paragraph IV sweet spot)
- 5-10 yrs to expiry: 0
- > 10 yrs to expiry: -2 (early-lifecycle, usually pillars)
5. Forward-citation adjustment (added May 2026)
Highly-cited patents are harder to invalidate. Citation count from USPTO PatentsView is enriched nightly via our citation pipeline.
- ≥100 forward citations: -10 (landmark patent)
- 30-99: -6
- 10-29: -3
- 0-1: +3 (rarely cited)
6. PTAB IPR outcome (added May 2026)
If the patent has been involved in IPR proceedings, the outcome is a direct signal. Source: USPTO PTAB E2E API, hourly refresh.
- Claims invalidated in final written decision: +25
- Petition instituted (not yet final): +12
- Petition filed but denied institution: -8 (validity signal)
7. Paragraph IV certification (added May 2026)
If at least one ANDA filer has staked a Paragraph IV certification against this patent, they've publicly bet it's vulnerable. Source: FDA monthly cert list, daily poll.
- Patent has ≥1 Para IV cert: +10
Combined score (v2 — May 2026)
score = base[classification]
+ age_adjustment(years_since_issuance)
+ family_adjustment(international_family_size)
+ runway_adjustment(years_to_expiry)
+ forward_citation_adjustment(USPTO_citation_count)
+ ipr_outcome_adjustment(PTAB_status)
+ paragraph_iv_adjustment(has_anda_filer)
score = clamp(score, 0, 100)
v1 (pre-May 2026): base + age + family + runway. v2 adds citations + IPR + Para IV — all derived from data we now refresh nightly via VPS scrapers. Backtested invalidation lift improved from ~4.8× to ~6.1× across top vs bottom score bands.
Validation against PTAB IPR outcomes
We backtest by pulling every PTAB IPR final decision on an Orange Book patent (2012-2025) where we have current scoring data. Findings (May 2026):
- Patents we score ≥ 70 (Vulnerable) were instituted at IPR and reached a final decision invalidating at least one challenged claim in ~58% of cases.
- Patents we score 0-24 (Ironclad) reached the same outcome in ~12% of cases.
- This is a ~4.8× lift in invalidation rate between top and bottom bands — meaningfully better than random and competitive with the major proprietary indices we benchmark against.
What the score is NOT
- It is not legal advice. Validation against historical outcomes is directional; specific challenges turn on prior art that we cannot inspect at scale.
- It does not account for filed-but-not-yet-decided IPRs. See the Biosimilar Opportunity Scanner or per-patent page for active IPR overlay.
- It does not distinguish between strong and weak instituted IPRs — only the outcome where one exists.
- It is not constant. Scores recompute daily as facts change. Major movements are surfaced via patent watch alerts.
Why we publish the formula
Pharma IP teams have been burned by black-box indices from incumbents. If you can't audit the reasoning, you can't trust the output, and you certainly can't defend a strategic decision to your GC or your board on the basis of it. Every parameter above is editable in source — open an issue or email us and we'll explain or revise.
Source code that computes scores: scripts/patents/compute-vulnerability.mjs in our repo. Database column: orange_book_patents.vulnerability_score.