Hello and welcome to Careviser by Marie Loubiere. Years of research summarized in 5-minute emails. Spot emerging opportunities in healthcare.
This week’s article is special. It evaluates the performance of a specific branded solution: Paincheck. We often doubt whether research published about a specific vendor solution can be trusted. It is key for vendors to publish scientific papers to show that their solution performs well. The methodology of such articles needs to be reviewed thoroughly to ensure reviewers were independent enough to not biaise the analysis and the results. For instance, when Natural Cycles published research in Contraception in 2017 to show that its app offered an efficient method of contraception, it was heavily criticized for its methodology (not a double-blind study, several researchers worked for Natural Cycles…).
Here, the study was funded by Paincheck. One of the authors is the Chief Scientific Officer of Paincheck and another one is a shareholder and co-inventor of one of their patents. The remaining one was paid by Paincheck to perform the biostatistical analysis. Now let’s focus on the study!
Assessing procedural pain in infants: a feasibility study evaluating a point-of-care mobile solution based on automated facial analysis, Kreshnik Hoti, Paola Teresa Chivers, Jeffery David Hughes, Lancet Digit Health 2021; 3: e623–34
🗝️ Why it matters: The management of pain associated with medical procedures (“procedural pain”) for children is a key priority for the World Health Organization. While adults and older children are able to self-report levels of pain, it’s obviously impossible for infants. Existing solutions are based on external observations of the child’s facial expression which are subject to bias.
🔎 The study: The objective was to assess whether the app Paincheck Infant could automatically detect pain in infants against two existing scales: the Neonatal Facial Coding System and the Observer administered Visual Analogue Scale.
The study relied on 40 videos of infants getting vaccinated which were divided into various segments before and after the vaccination.
Paincheck uses AI to automatically detect 6 facial expressions indicative of pain: brow lowering, wrinkling of the nose, lip corner depressor, horizontal mouth stretch, parting lips, and eye closure and indicates whether they are present or not with a binary rating. The maximum score is thus 6.
The Neonatal Facial Coding System relies on 5 facial expressions with a rating not done automatically, but by a reviewer.
The Observer administered Visual Analogue Scale is a visual scale where reviewers estimate from the distance to 0 the level of pain experienced by an infant.
2 reviewers (one trained and one untrained in pediatrics) reviewed each video segment and rated them according to the Neonatal Facial Coding System and the Visual Analogue Scale while the Paincheck algorithms performed their own rating automatically.
✅ Findings: Ratings by Paincheck were consistent with the Neonatal Facial Coding System and the Observer administered Visual Analogue Scale (beware none of these existing scales are considered as gold standard).
🚀 Opportunities ahead: The study was done using recorded video. The question of Paincheck clinical utility in real life remains. The provider would need to use a smartphone to capture the child's facial expression. They may need to use their hands at the same time (e.g., to vaccinate the child) which could make the user-friendliness of the system tricky. There won't always be another provider or a nurse available only to hold a phone and use an app.
A bit more about Paincheck.
Paincheck started out as an app for providers in care home facilities. It was a spinoff of existing research in a university lab. The research started in 2012, and by 2016, they were acquired by an existing company: MinQuest. MinQuest was a small public company active in the mining discovery field. They had IPO’ed on the Australian Stock Exchange (with a small market cap of a few million AU$) after several years of existence. I guess they were not that successful with their mining ventures, and spotted Paincheck, decided to acquire them, and help them go to market (so random I know!). The former chairman of MinQuest is still chairman at Paincheck (he is a PE veteran in Australia).
The Paincheck app has been live since 2017 and they have received regulatory clearance as a medical device in the US, Europe, and Australia for the non-verbal adults use case.
Old patients cannot always verbalize pain levels especially the ones suffering from dementia. Paincheck has more than 60% of Australian residential aged care facilities as clients. Their go-to-market was supported by a government grant: Australian residential aged care facilities could get a free trial for a year fully covered until May this year. They report that ARR went from AU$2m in 2019, to AU$3m in 2020 and AU$6m now. ARR assumptions for 2021 could be challenged. They are based on no-churn for clients that entered the program through the grant when historically 25% of them dropped out. Still, their accomplishments in a few years are impressive, especially as their team is small (<30 people). I actually wonder how they built such a product with only a few software engineers on the payroll.
Paincheck now aims to go after two new segments:
Home care providers who travel to the patient’s home. A lot of these providers also work in care homes so they are already aware of the solution making their go-to-market easier.
Providers in the pediatric space (vaccination, surgery…) when they get their kids' use-cases cleared by regulators.
They are also expanding internationally starting with the UK.
Despite all the progress, Paincheck’s share value has not increased in the past years. So I’ll be keeping an eye on them.
That’s a wrap for today! Don’t hesitate to reply to this email with comments, I read and answer all emails :)