Neurosymbolic Loon AI® · Peer-reviewed · Independently validated

Confidence-calibrated AI systematic literature review

Loon Lens™ automates the systematic review process — from deduplication, study classification, and screening through study mapping, data extraction, and quality assessment — with calibrated confidence that routes the rare uncertain decision to human review.

See the evidence

Evidence that lands.

99.0%SensitivityTitle & abstract screening with ~5% confidence-guided review — Value in Health (ISPOR), 2025 99.1%SpecificityWith confidence-guided review of just ~5% of records — Value in Health (ISPOR), 2025 100%Extraction truthfulnessIndependent Cochrane benchmark — PAS, 2026 98.9%Workload reductionValidated independently by the BC Neonatal Cochrane group

Systematic reviews are slow, costly, and error-prone.The evidence base doubles every 73 days.

The average systematic review costs $141K and takes 67 weeks. Expert extraction covers only 50–54% of baseline data, and 85% of published SLRs contain extraction errors — despite 9 million person-hours spent on reviews every year.

Loon Lens™ compresses this cycle: peer-reviewed, confidence-calibrated agentic AI that takes teams from eligibility criteria to an audit-ready, analysis-ready evidence base — in days, not years.

The platform

One platform from search through screening, mapping, and extraction

Loon Lens™

Confidence-calibrated
AI systematic lit reviews

Just assess eligibility criteria and upload citations — no pre-labelling, no pre-training, no fine-tuning. Specialised Loon AI® agents deduplicate, enrich, and classify on ingest, then screen, reason, challenge, and adjudicate decisions independently.

One-shot extraction from tables, figures, and text — harmonised across studies, no caps
Automatic PRISMA 2020 flowchart, study mapping, and per-datapoint provenance
Any study design or review type — treatment effects, prognostic, cost-effectiveness, qualitative

99.0% sensitivity99.1% specificity100% extraction truthfulness

Loon Lens™ — confidence-calibrated AI systematic literature review, from eligibility criteria to data extraction

How it works

From eligibility criteria to audit-ready evidence

You bring the eligibility criteria and the citations. Specialised Loon AI® agents do the rest — reasoning, challenging, and deliberating with full provenance at every step.

01 · InputWhat you bring

Eligibility criteria (PICO / PEO)
Citation records from any database
Optionally, specific variables to extract — by default, we extract everything

That is the entire setup — no pre-labelling, no fine-tuning.

02 · DeliberationWhat Loon AI® does

Deduplicates, enriches, and classifies citations on ingest
Screens titles, abstracts, and full text — multi-agent
Maps multiple reports to their main studies
Auto-generates the data schema and extracts every datapoint from tables, figures, and text — one shot, unrestricted by number of studies, columns, or variables
Adjudicates disagreements and routes low-confidence items to human review

03 · OutputWhat you receive

Harmonised, analysis-ready dataset — multi-sheet Excel, R/Python-ready
Dynamic PRISMA flowchart and automated study mapping
Per-datapoint provenance with the exact source quote
Calibrated confidence on every decision and value extracted
Full audit trail and QC with Loon AI® reasoning per decision

Compress the evidence cycle. From months of manual screening to a single pass — fully autonomous or with human-at-the-helm confidence-routed validation that sends only the rare uncertain decision to an expert reviewer.

The complete evidence lifecycle

Loon Lens™Search & screenFrom eligibility criteria and citations to screened evidence

Loon Lens™Extract & synthesiseOne clean, harmonised dataset with full provenance

Loon Waters™ · loonbio.comSimulate & forecastPayer-ready value evidence and de-risked submissions

Together, Loon Lens™ and Loon Waters™ — Loon's reimbursement-forecasting platform — take teams from literature search through to payer-ready value evidence, compressing the entire evidence cycle.

Calibrated confidence

The only AI that knows when it is uncertain

Every Loon AI® decision carries a calibrated confidence score. Uncertain decisions — typically under 6% — are automatically routed to targeted human review, so teams review only what needs review, not everything or nothing.

Confidence-guided review of just ~5% of records raised precision from 63% to 89.9% and specificity to 99.1%, with 99.0% sensitivity and 99.0% accuracy — validated across 3,796 citations from 8 CDA-AMC reviews.

Specialised agentsScreen, reason, challenge, extract, and adjudicate decisions independently

Mathematical guardrailsNeurosymbolic architecture ensures reliability and full decision transparency

Flexible workflowsAI + human, AI-only, or human-only — masked or open review frameworks

Services

AI-enabled services for HEOR, HTA & market access

Alongside the platform, Loon's team of epidemiologists, HEOR scientists, and evidence-synthesis experts delivers full engagements — as SaaS, tech-enabled services, or fully traditional service, whichever fits your workflow.

Evidence synthesis: SLR, LSR, ITCPeer-reviewed agentic AI conducting HTA-grade (living) systematic literature reviews, meta-analyses, and ITCs (NMA, MAIC, ML-NMR) with efficiency.

ITC feasibility & network evidenceDefensible feasibility studies and NMAs quantifying relative effectiveness when head-to-head evidence is limited — HTA- and JCA-grade evidence at speed.

EU JCAs support & PICOs executionRapid HTA-grade living SLR/LSR® — search strategy, screening, extraction, and ITC feasibility for every PICO in scope, with eligibility criteria stress-tested before they land in the dossier.

Scientific validation

Peer-reviewed. Independently benchmarked.

Pediatric Academic Societies · 2026 · PresentedCan agentic AI replicate a Cochrane neonatal systematic review?Independently benchmarked by the BC Neonatal Cochrane group across 1,865 records: 98.9% workload reduction, 94.1% study concordance, 100% data extraction truthfulness.Mitra, Fisher-Huise, et al. · 2026.pas-meeting.org Value in Health (ISPOR) · 2025 · Peer-reviewedValidating Loon Lens 1.0 for autonomous abstract screening and confidence-guided human-in-the-loop workflows98.9% sensitivity and 95.2% specificity in full AI-only mode across 3,796 citations — rising to 99.0% sensitivity, 99.0% accuracy, and 99.1% specificity with confidence-guided review of ~5% of records, exceeding dual-human screening without pre-labelled training data.Janoudi G, et al. · doi.org/10.1016/j.jval.2025.09.008 medRxiv · 2025 · Pre-printEvaluating Loon Lens Pro, an AI-driven tool for full-text screening in systematic reviewsA validation study extending autonomous agentic screening beyond titles and abstracts to full-text review — the stage where most expert screening hours are spent.medrxiv.org · 10.1101/2025.02.11.25322087

Loon at ISPOR · Short courseApplied Generative AI for HEOR: Advanced ArchitecturesLoon leads ISPOR's advanced short course on generative AI architectures for HEOR.

Loon at ISPOR · Issue panel, 2025AI Agents and Guardrails in HEORInvited panellist alongside ICER and Bristol Myers Squibb, in a standing-room-only session on the trustworthiness of agentic AI.

Loon at ISPOR · Posters & papersPeer-reviewed contributions across the programmeIncluding autonomous data extraction from clinical study reports, benchmarked against Canada's Drug Agency reports.

All publications

Who uses Loon Lens™

Built for every team that values evidence

Pharmaceutical sponsorsAccelerate SLR timelines and accuracy, generate HTA-compliant evidence syntheses, and free expert teams for submission strategy and interpretation — not screening and extracting.

HTA bodies & payersReview submitted evidence more efficiently with transparent, auditable AI-assisted workflows — confidence-calibrated and fully traceable to source.

CROs & consultanciesDeliver more reviews in less time without growing headcount — peer-reviewed methodological rigour at scale, for every client engagement.

HEOR & clinical research teamsBuild comprehensive evidence bases for value dossiers in days, refresh living reviews continuously, and harmonise extraction across studies into one clean dataset.

Security & compliance:No training on client dataZero-day retention with all LLM providersGDPR & PIPEDA compliantFAIR & RAISE audit trailsConfigurable data residencyCompliant with ISPOR ELEVATE-AI, NICE & CDA-AMC position statements

Clients

Our clients become our biggest champions

We work with HTA bodies, Fortune 10 healthcare companies, consultancies, and independent researchers alike — and the HEOR professionals who use Loon are the ones who recommend us.