The mental health prognosis in yourpocket

Original article: nbr.co.nz

Sahha co-founder Aleks Dahlberg

Imagine an app on your phone that monitors your mood – pinging you a notification when it senses you’re depressed or anxious or burnt-out, and triggers an intervention to help address the imbalance.

The founders of Sahha – Aleks Dahlberg and Doug MacDonald – are trying to make this a reality, and while it all sounds a bit ‘big brother’, they are certain the technology can be developed in a way that is useful and ethical for both individuals and organisations. The word Sahha represents health, welfare, and strength in Arabic, according to the company.

They’ve managed to convince about $1 million-worth of backers of their vision so far: Sahha had A$190,000 pre-seed funding on its path through Australia’s Antler accelerator where it was founded in early 2021, and where the pair met.

They have since sealed another $800,000 in funds from Mark Pavlyukovskyy’s NZVC fund, Blackbird Ventures, “a few angel investors from Australia and one from the US”, as well as Antler again.

Dunedin-based Dahlberg told NBR his goal is to “have optics” on a million dollars ofannual recurring revenue in the next 12 months. He also wants “optics” on a Series A raise in the same time period, but is currently seeking another $2m to tide the company over.

But let’s not get ahead of ourselves. What’s all this money for? At its root, Sahha’s role is to create code that accurately identifies certain human behaviours from even the most innocuous data captured by an on-person device.

As people become more comfortable with using smartphone and wearable technologies to monitor things like their footstep count, heart rate and sleep patterns via health and wellbeing apps, a natural next step is for the apps to make judgments based on more than just targets assessed by the user themselves.

Cutting self-assessment from this process reduces the risk that the assessment can be falsified by the subject, said Dahlberg, and that’s particularly helpful in a mental health setting when people may not be honest with themselves.

How does it work?

Sahha doesn’t create apps though – there are plenty of those out in the market already: Apple’s Health app for instance, which in-turn uses sensor data stored in the HealthKit app.

In practice, Sahha’s technology is embedded into these kinds of third party apps via an API (application programming interface) or SDK (software development kit) which then feeds the data back to Sahha’s system – anonymously – and runs an “inference” on the data. It uses the data to predict a condition, and sends the result back as a piece of code to the third party.

Sahha never receives any identifying information through that process, and the rest is in the hands – and permissions – of the app developer and the end user.

A screen capture from Sahha's website showing one of its prognoses.

Many apps are already privy to the kind of data Sahha uses, but they’re not making the most of it, Dahlberg said.

Through the third-party app, Sahha’s finding might prompt something as simple as a notification to the user, or a referral to a care programme or telehealth session.

It might even display the result visually on an enterprise wellbeing dashboard so that managers can monitor productivity and performance levels of staff to satisfy promises made to stakeholders.

The range of prospective customers is broad, noted Dahlberg, including everyone from primary care providers and GPs through to chronic care and remote patient monitoring, as well as employee assistance programmes.

“Over in the States, there's reinsurers... that are trying to provide care for people. By being more preventative, the industry ends up making more money on the insurance claims. There’s so many different use cases here, it's incredible.

“Using mental state data, we have the power to connect a lot of very broken healthcare industries together.” NZVC’s Pavlyukovskyy told NBR he was excited about the possibility of the likes of Apple Watch or Fitbit integrating Sahha’s technology into its existing platform. It could very well be a free feature for them to offer, but it’s likely to create a lot of value for third parties.

His fund has invested in two other mental health related businesses recently too – employee wellbeing reporting app Chnnl and VR phobia therapy service Ovrcome – and is a firm believer that technology can solve some of the great health issues of our time.

NZVC founder and Edmund Hillary Fellowship recipient Mark Pavlyukovskyy

Product development

Much of Sahha’s technology is still in Alpha or Beta mode, and Dahlberg admits the information Sahha generates is not a diagnosis, but it can still be extremely useful.

He also believes it could one day get to the level where it can be clinically recognised as a diagnostic tool.

“We call it a prognosis, if anything, and... personally, I think it will outperform a clinician, for sure. It can outperform self-assessment, because self-assessment is incredibly biased, subjective and subject to all sorts of environmental, situational and time-related factors.”

So where’s the proof so far?

Dahlberg said the company has about 10 research trials ongoing with its research partner, involving more than 3000 research participants from the UK, EU, the US and New Zealand. Some of the trial participants have been on-board for 30 or 40 weeks, meaning the company has tonnes of data with which to train its models.

Sahha’s team of around 10 engineers and data scientists are adept at picking up flaws in the way other companies and academics in the space are conducting their research, Dahlberg said, so he expects the company will be publishing its own academic papers over the next year – showing the world how much better its models are.

Dahlberg said he “can’t go too much into the secret sauce” that sets Sahha apart from its competitors, but that it's “significantly different than just comparing behavioural data or psychometric instruments and trying to generate a model on that level.

“A lot of this is around how you think about solving the problem and not necessarily about...,” Dahlberg halted before giving too much away. “Yeah, I think it's just about how you approach the problem solving.”

Meanwhile, the commercialisation continues at pace, he said.

“We have models that are tight now. They're very MVP, they’re very Alpha-kind-of models and we're hoping to move those to a Beta state in literally like two weeks.

“But the research thing for us is ongoing. We always want to be improving, we want our initial models to be improving and getting better and better for our customers.

“Validating everything on the data science side of things, we want that to improve more as well: how we clean data, how we look at data, the feature sets that we can extrapolate from the data points, the machine learning models that we use – all these things where we're intensely looking at all the time.”

Contact the Writer: william@nbr.co.nz