C-SATS

People perform the assessment and machine learning supports it. The How it Works page says case videos receive qualitative feedback from expert surgeons and quantitative reviews from crowd-sourced reviewers using post-operative assessment tools, with scores combining crowd inputs as weighted averages; the name, Crowd Sourced Assessment of Technical Skills, describes the method. The Enterprise and pricing pages list the machine learning around it: AI-assisted removal of patient information from video, step segmentation for select procedures within one to two business days, and AI insights and analytics on the surgeon's dashboard.

Those functions are real and useful, and what an institution buys is expert and crowd review of its surgeons' video at scale, a reference library of more than 19,000 procedure videos and a surgeon community, which stand without the models. The company markets the combination as AI plus human insight rather than presenting the human review as AI scoring.

People make the judgment about a surgeon, and the AI steps around them are not bounded in public. Scores come from crowd reviewers and expert surgeons applying published instruments; feedback returns to the surgeon's own dashboard; video sharing stays within the institution and, in some instances, the C-SATS community; and the How it Works page says health systems can see trends and performance across the surgical team. Enterprise licenses cap expert reviews at two a month and quantitative scores at twelve a year per license.

What is not published is whether a person checks the AI removal of patient information before a crowd reviewer sees the video, what the AI insights on the dashboard decide or recommend, and what institutional leaders see about an individual surgeon.

The method that produces scores is published in full, and the machine learning is named only by task. The help center publishes the GEARS and GOALS instruments with each domain's 1 to 5 anchors and total ranges, and the How it Works page says quantitative scores are weighted averages of crowd inputs, with qualitative reviews by expert surgeons.

The AI functions are named (removal of patient information from video, step segmentation for select procedures, dashboard insights) without the models behind them, their versions or how they are updated.

The host and processor are named, and the model layer is not. The Data Security page names Amazon Web Services in Oregon as the host and a Johnson and Johnson affiliate as the processor, and describes the recipients of video: crowd reviewers and expert surgeons (some named on the site as C-SATS experts), the surgeon's own institution and, in some instances, the C-SATS community.

The site's global privacy policy allows disclosure to affiliates and combining information with that from other Johnson and Johnson affiliates. Nothing says who builds or runs the models that remove patient information and segment procedures, whether a third party is involved, or what reviewers may retain.

Published research on the assessment method sits behind the product; the outcome claim rests on an unpublished analysis. The company's publications library lists peer reviewed work on crowd-sourced assessment, including JAMA Surgery papers and a 2022 systematic review of the platform's evidence, and names deployments through dated releases (Northwell and Providence St. Joseph Health in 2017, AdventHealth, Hackensack University Medical Center in 2020, Valley Health System, UC Irvine urology, the University of Washington and a US Air Force surgeon program). Customer quotes from Hackensack Meridian Health and AdventHealth describe improvements without figures.

The home page's headline figures come from what the company labels an unpublished exploratory analysis: about 3,000 cases across six specialties and 25 procedure groups, comparing surgeons in the top and bottom quartiles of GEARS scores, with confidence intervals and no adjustment for multiple testing. That shows skill scores track outcomes; it does not show that using the platform improves them, which is what the company says it may do.

Training use and de-identification are stated specifically; retention is not. The Data Security page lists what is collected (surgeon names, procedure type, date and time stamps, video from the laparoscopic tower or robot, optional notes and case context such as gender and BMI), says out-of-body footage is removed before video is shared for assessment, and says data may help improve AI and machine learning models only under an optional, separate agreement; the pricing and Enterprise pages add AI-assisted removal of patient information, noting that metadata may include date and time stamps. The hospital or surgeon is the data controller and a Johnson and Johnson affiliate the processor, under a business associate agreement.

No retention period is published for video, scores or metadata, and no accuracy is published for the automated removal of patient information. Safety engineering, meaning how a failed removal or a wrong AI insight is caught and handled, is not published either.

Business associate status is stated, with the tier it applies at, and the agreement is not published. The pricing and Enterprise pages say a signed business associate agreement is required to activate Enterprise services, alongside a customer service agreement and scope of work, and the Data Security page says the hospital or surgeon is the data controller, a Johnson and Johnson affiliate is the processor, and a business associate agreement is in place between the parties; HIPAA, HITECH and GDPR are named as the laws the platform is designed around.

The agreement itself is not available to read, so a buyer cannot see whether it binds the crowd reviewers and expert surgeons who view de-identified video, which is the question that matters here.

A recognized certification is named without the artifact. The product pages and Data Security page say the case library is HITRUST CSF Certified, and name HIPAA, HITECH and GDPR as the laws the platform is designed around; the certificate, its date and scope are not published, and there is no trust portal. The AWS certifications the page lists (ISO 27001, HITRUST, SOC 2 Type 2) belong to the host, not to C-SATS. The site footer reads 2022, so the currency of the certification is a question to ask rather than assume.

The company states its position and the product matches it. Every page footer says the C-SATS product is not a medical device, is not intended to replace formal medical education or training, is designed for licensed surgeons, and does not provide medical advice; access requires an active National Provider Identifier. What is sold, review of surgeons' recorded technique with feedback and benchmarks, fits that statement. The company does not publish its reasoning for sitting outside device regulation.

The exposure that does apply is legal rather than regulatory: scored video of a named surgeon's operation is the kind of record a plaintiff would seek, and whether it is protected depends on state peer review privilege, which was not written with outside reviewers in mind.

The method is documented; its fairness is not evaluated. Scoring uses published instruments (GEARS for robotic and GOALS for laparoscopic surgery, each domain on a 1 to 5 scale, with the scoring anchors reproduced on the help center page), and the publications library lists validation studies of crowd-sourced assessment, such as whether crowd scores track surgeon experience in robotic bariatric surgery. Nothing on the company's pages reports inter-rater reliability, whether scores vary by surgeon characteristics unrelated to skill, or how crowd reviewers are selected and calibrated, and the AI functions (patient information removal, segmentation, insights) carry no published testing.

The platform belongs to Johnson and Johnson MedTech, which sells surgical robotics and instruments, and no statement says assessment and benchmarks are neutral across device platforms. Video capture is described as platform agnostic.

The basis of every score is published, and nothing stands behind it. A surgeon's quantitative score comes from published instruments whose scoring anchors appear on the help center page, combined from crowd reviews as weighted averages, with qualitative feedback from named expert surgeons; a surgeon can read the criteria and argue with a score on the instrument's own terms. The terms of use make the service available as is, disclaim warranties and say the company gives no assurance about the accuracy of the information provided.

No route to dispute a score or ask for re-review is published, and no accuracy or error rate is given for the AI removal of patient information from video, which matters because the video then goes to reviewers outside the institution. The pricing page notes that metadata may still include date and time stamps.

The product takes clinical record data, and no record system is named. The pricing and Enterprise pages say outcomes analytics requires access to the customer's EHR data, and the home page reports outcomes such as length of stay and blood loss drawn from EMR data; no EHR, interface standard or direction of exchange is named.

Video capture is described as agnostic to minimally invasive surgery platforms, by C-SATS hardware or manual upload, which is the other integration that matters to an operating room, and no capture system is named either.

The data location is committed. The Data Security page says the platform is hosted on Amazon Web Services in a data center in Oregon, United States, with data encrypted in transit and at rest, and that EU Standard Contractual Clauses cover transfers of data from outside the EU so surgeons abroad can take part. Each institution has its own case library; video can be shared within the institution and in some instances with the wider C-SATS community.

What is left open is the processing path for review: crowd reviewers and expert surgeons, including reviewers outside the United States, view video through the platform, and nothing says what they can download or retain.

One tier is priced and the paid tier is not. The plans and pricing page lists Basic at $0 per month, open to any surgeon with an active National Provider Identifier through online sign-up (surgical video academy and surgeon community), and Enterprise at contact for pricing, with its features and limits listed: automated capture via C-SATS hardware or manual upload, AI-assisted removal of patient information, secure case library, step segmentation, analytics and benchmarks, AI insights, up to two expert qualitative reviews a month and up to twelve quantitative scores a year per license, outcomes analytics, and a dedicated support team. Enterprise needs a signed customer service agreement, business associate agreement and scope of work, and the plans apply in the United States only. The home page also offers Premium and Advanced memberships, which the pricing page does not show.

The unit for the paid product is the license, with its review allowances stated, but no price.

Coverage is named within surgery, with published work behind part of it. The platform assesses robotic and laparoscopic surgery with the GEARS and GOALS instruments, and its video library spans more than ten specialties including bariatric, gynecology, urology, thoracic and colorectal; the home page's outcomes analysis covers gynecology, general surgery, urology, urogynecology, gynecologic oncology and colorectal. Dated releases name use by practicing surgeons (Northwell, Providence St. Joseph Health, AdventHealth, Hackensack), residency programs (University of Washington, Valley Health System), applicant evaluation (UC Irvine urology) and military surgeons, and plans are offered in the United States only.

The validation work covers some procedures and settings and not others, and the product serves surgeons rather than the wider workforce.