Dandelion Health

1. An organization requires publicly documented SOC 2 or ISO certifications prior to procurement. The vendor may hold such certifications privately, but they are not publicly documented in the sources reviewed.
2. The primary need is claims-based patient journey analytics and no images, waveforms, or notes are required. Other vendors with claims-centric datasets may be a more direct fit.
3. The requirement is a proven, category-leading depth in oncology evidence. Established oncology RWE vendors are often the default benchmark in that niche.
4. The procurement process requires a large public review footprint (G2/Capterra/TrustRadius) to de-risk purchase. This vendor is underrepresented on those platforms.
5. The organization requires immediate, daily-refresh national-scale EHR data across many domains. Other vendors emphasize daily updates and larger national coverage.

The vendor launched a GLP-1 Data Library described as a multimodal dataset for GLP-1 receptor agonists:

• Press material states the library contains a large cohort of treated individuals and matched controls, with periodic refreshes and sourcing from multiple health systems.

Critical evaluation:

• These metrics are vendor-published; methodology and inclusion criteria are not fully disclosed publicly.
• Reported coverage and counts require clarification: are these unique patient records, lives covered by contributing health systems, or longitudinal records meeting completeness thresholds?

Vendor announcements state results indicating a relative risk reduction in major adverse cardiovascular events for GLP-1 users versus comparable non-users over follow-up and include population-level extrapolations to broader at-risk groups.

Critical evaluation:

• These appear to be observational, vendor-conducted analyses; causal inference limitations (confounding, selection bias, adherence, indication bias) should be assumed unless full methods and sensitivity analyses are published.
• The translation from relative risk reduction to absolute event reductions is highly assumption-dependent (uptake, adherence, baseline event rates). Methodology transparency is necessary for decision-grade use.

The vendor published a report, funded by a foundation, describing a validation framework evaluating model performance against social determinants of health and demographic measures, with findings that include potential underperformance for older adults and rural populations.

Critical evaluation:

• This aligns with known market needs for bias detection and representativeness assessment.
• Without independent peer review of the framework, procurement teams should evaluate how it compares with internal model governance practices and competitor offerings.

Dandelion Health is a B2B real-world data (RWD) and clinical AI platform that provides multimodal, longitudinal patient records intended to support evidence generation and clinical AI development for regulatory and research purposes. The vendor describes access to linked structured EHR data plus unstructured modalities such as clinical notes, medical imaging, and physiologic waveforms, and describes a Clinical AI Marketplace of algorithms used to convert unstructured data into structured, research-oriented labels.

Key platform pillars described by the vendor:

• Dandelion Data / Precision Medicine Platform: multimodal longitudinal patient data sourced from a consortium of health systems; vendor describes this data as suitable for regulatory submissions and research use.
• Clinical AI Marketplace: a marketplace that the vendor states hosts algorithms to structure notes, images, and waveforms into labels/features for research and development use.
• Use cases across the product lifecycle: early clinical insights, clinical development (cohort selection / trial planning), post-approval evidence, indication expansion, and AI developer workflows (training, validation/bias checks, regulatory support, commercialization).

Primary buyers/users appear to be:

• Life sciences (pharma/biotech) teams responsible for RWE/HEOR, label expansion, trial design, and post-market evidence.
• AI developers, medical device manufacturers, and digital health companies needing training and validation datasets for clinical algorithms.

The vendor differentiates itself from some traditional RWD providers by emphasizing multimodal data (notes, images, waveforms in addition to structured EHR data) and by framing its provenance and validation practices as suitable for regulatory use; these are vendor claims and should be validated during procurement.