Defining AI biomarker scoring and why it matters now
AI biomarker scoring is the use of artificial intelligence to detect, quantify, and standardize cancer biomarkers on digital pathology slides, so that treatment decisions rely on consistent, reproducible measurements rather than subjective estimates from manual review alone. That shift is moving from research labs into clinical pathology AI workflows as imaging vendors and AI specialists align on shared infrastructure and validation paths. The promise is clear: for biomarkers like PD-L1, which guide access to targeted therapies, even small differences in scoring can change patient eligibility and treatment order. By embedding algorithms directly into cleared digital pathology workflows, companies aim to reduce inter-observer variability, speed up reporting, and create an audit trail suitable for regulators and payers. The transition now underway is less about proving that algorithms can work, and more about proving that they can work reliably, at scale, in routine practice.
Leica, Indica Labs and Lunit turn PD-L1 detection into a product
Leica Biosystems, Indica Labs, and Lunit have turned years of AI biomarker scoring research into a production-ready workflow focused on PD-L1 detection. Their first joint algorithm, Lunit SCOPE PD-L1 CAL10 NSCLC, is built specifically for Leica’s PD-L1 primary antibody (CAL10) and is already live in the Aperio AI Store. It links Leica’s Aperio GT 450 scanner with Indica’s browser-based Aperio HALO AP DX image-management software and Lunit’s SCOPE algorithm, forming a single end-to-end workflow for PD-L1 scoring. According to Leica’s Karan Arora, “What AI does is drive consistency and put a quantitative layer on top of the expert pathology review,” shifting the pathologist’s role from manual estimation to confirmation of a quantitative output. Moving this kind of clinical pathology AI from pilot to deployed app store listing signals that AI scoring can now be delivered as a standardized component of routine diagnostic workflows.
Standardized workflows as the backbone for clinical pathology AI
The Leica–Indica–Lunit collaboration shows how AI biomarker scoring depends on tightly controlled, yet open, workflows. Leica has paired its Aperio GT 450 DX scanner with a clinically validated version of Indica’s HALO AP platform, distributing it exclusively for clinical use. On top of this stack sits the Aperio AI Store, which allows third-party algorithms from partners such as Lunit, MindPeak, pharma developers, or large labs to plug in through software development kits and APIs. The goal is to scan a slide once and interpret it the same way everywhere, with AI providing a quantitative PD-L1 detection readout that pathologists review rather than recreate. This architecture helps standardize pre-analytical steps, imaging, and analysis, while still allowing a best-of-breed AI ecosystem. It also creates a natural channel for ongoing validation, updates, and version control, all of which are essential for long-term cancer biomarker verification in regulated environments.
Verification as a new infrastructure layer for AI diagnostics
As AI biomarker scoring moves into clinical use, verification is emerging as a distinct infrastructure layer rather than a feature of any single tool. Aimwell Partners, through its AimwellBio healthcare intelligence network, is exploring how AI-generated insights can travel with their own proof trail. Under its proposed model, when a disease signal or expert insight appears, the system could show who submitted it, when it changed, who reviewed it, and what evidence supports it. The company is evaluating tamper-evident timestamps, provenance tracking, secure research vaults, and contributor verification. That approach echoes the needs of clinical pathology AI, where PD-L1 detection outputs must be traceable and auditable across sites and over time. As Aimwell notes, “Every major platform adding AI is, without intending to, expanding the market for verification,” suggesting that the next competitive edge will be not only smarter algorithms, but clearer proof of how they were produced.
Regulatory and reimbursement pathways for AI biomarker verification
Regulators and payers are pushing AI pathology from experimental promise toward accountable practice, and biomarker verification is becoming a key requirement. Leica’s FDA-cleared digital pathology workflow provides a regulated framework into which AI biomarker scoring tools can be embedded, allowing PD-L1 detection algorithms to operate inside an approved pipeline instead of as standalone research add-ons. Verification layers, such as those AimwellBio is exploring, could complement this by documenting algorithm versions, evidence sources, and validation histories when outputs influence treatment choices. For cancer biomarker verification to scale, multi-vendor partnerships must also support clear traceability for updates, consistent performance across scanners and labs, and transparent evidence packages for regulators. In this model, AI biomarker scoring is not only a new analytical capability; it becomes part of a broader, provable system that links scan, algorithm, pathologist review, and final report into a single, inspectable chain of clinical responsibility.
