Download the full Insight Project article here.
Inferring Complex Data from Consumer Biometric Devices
Using scientific measurements as the basis for widely applicable purposes confronts us with a paradox. The essence of measurement is precision and reliability which can entail expensive, bulky and temperamental equipment set-ups. But, while we wish to optimise precision and must therefore use high-end equipment and rich complex datasets, we also require easy applicability in a wide range of settings. To begin with, the science will take precedence, but we must then turn our attention to fashioning practical devices, affordable and usable by the ordinary person.
In this respect, everyday devices such as Fitbits and phones are able to track biometric data and we may be able to capitalise on these. Alternatively, it might be possible to create and manufacture a custom device that fits our needs. However, the data is not going to be anywhere close to the precision of specialist medical devices nor will it track the range of data we can measure in our lab.
How do we bridge the gap? We anticipate that a machine learning system could be trained using the high-quality lab hardware and a large enough sample size. Its ensuing application to more sparse datasets would be optimised by the prior principles and knowledge gained from a growing database.
To achieve this a significant R&D effort would be required and access to machine learning technology and expertise.
Employ machine learning to estimate high-quality data from sparse biometric data found in mainstream consumer devices
Optimising for Success with Data Analytics
In traditional experimental work in neuroscience, small, sparse datasets – often collected in laborious ways – place a limit on progress. But it is common practise in games, especially in online games-as-a-service, to track many datapoints for each player and across all players and thereby to acquire extraordinarily rich longitudinal measures of engagement and performance. This data is typically used to track progress and engagement and to optimise monetisation. We see an alternative potential for this precise and intensive, but simultaneously easy and unobtrusive, monitoring.
From this data, global changes to the game are made over a period of months and years to improve players’ experiences and, crucially, the data is fed back to each individual player to offer a tailored experience that encourages engagement and a measured sense of progress and success.
It is likely that big data analysis and machine learning techniques will have to be developed as the data set grows. The more data that is available, the more precise we can be about optimising for success. Of course, this presents challenges with respect to individual privacy and we will commit to ensuring that GDPR considerations are at the forefront of any such developments.
Data analytics using a large sample size of players can help optimise each of us gain expertise in biometric control
Transferring Skills from the Virtual to the Real World
If we reach this point, we will have created a rich and rewarding experience to help people to manage, control and reframe their fear in a controlled and curated simulation. But how to carry over this skill into everyday life?
In many ways, this challenge is akin to those faced by the traditional therapist whose goal is to help a person take insights from the controlled therapeutic setting and apply them more generally. We can therefore learn from how these challenges have been met in the past. For example, a therapist may give a patient a stimulus or memento that can provide a focus during times of stress or panic and thereby prompts the use of learned techniques. In combination with increasing feasibility of continuous monitoring of key individual variables (such as heartbeat), this approach could prove highly effective.
Another observation is that players who become experts at a game, require fewer and fewer feedback elements to be able to successfully continue playing that game, as is demonstrated in this video of a Tetris player playing with invisible shapes:
As players progress through the Insight Project, we can design the game to rely on fewer and fewer feedback elements. It may be that over time, you can reduce the biometric feedback elements to zero, simply by virtue of the player having become aware of their self through sustained play.
More likely, perhaps a wearable device such as a ring, watch or band can offer vibrational or audio cues to help players in everyday life cope with a situation and guide them back to calm.
We will find ways for learnt techniques to carry over into everyday life with minimal intrusion
Clinical Application & Knowledge Sharing
Since the release of Hellblade, the game has received some interest from clinicians and researchers where it has been employed as an educational tool and as the basis for limited scientific study.
For the Insight Project, we can design the experience to offer a far more sophisticated set of tools that would allow researchers to modify, test and publish scientific studies. There is precedence for this in game development circles in the form of “modding”.
Modding is a common gaming term whereby a game is released with an allowance for sophisticated user-customisation. This is typically used to create variations of the experience or, in some cases, entirely new game experiences. Combined with our commitment to GDPR and the Open Science framework, we can offer the scientific and clinical community a tool set and data for further research, and to clinicians, a platform for exploring new therapeutic strategies.
There are several advantages to this open approach: it will encourage more research and knowledge sharing that could feed back into the Insight Project resulting in a more effective product or service; it will help encourage scientific validation of the Insight Project from independent sources; and last but not least, it will help spread new ideas and approaches to mental health treatment.
We will aim to provide a platform for further research and clinical application to help inspire the Insight Project as a legitimate approach to mental health