I’m involved in a new EU project called PhilHumans, which starts on 1 Jan 2019. The project will support 8 PhD researchers around Europe, who will work with Philips on using AI techniques to improve how users interact with personal health apps. Its a really exciting project, and I encourage potential PhD researchers to get in touch with me!
[Update 11-Jan-19] You can formally apply for the ESR7 position on the Aberdeen recruitment page.
Personal Health Apps
Our actions have a huge impact on our health, and the health of our families. Almost everyone will benefit from eating sensibly, exercising, and driving carefully, not to mention brushing teeth properly. And many people need to manage specific medical issues. For example, someone with asthma (like me) needs to know when to use an inhaler, when to see a GP, and when to go to the hospital A&E.
There is huge interest in tools which help people manage and improve their health; the interest comes from health systems and healthcare providers as well as individuals. I think that AI and NLP can play a very important role, especially in personal health “apps”, by helping people understand their health and the (health) consequences of their actions, and also by helping people change their behaviour (if they want to change!).
I’ve been involved in a number of projects in this area in the past, including systems which encourage smoking cessation and safer driving, and a system which keep parents informed about the status of sick babies in hospital. In terms of current projects (in addition to PhilHumans), I am currently supervising one student working on safer driving, and also am assisting in a project to encourage elderly people to exercise. I also have funding (with the Aberdeen Medical School) for another PhD student in this area, who will start in summer 2020. I may have additional funding later in 2019.
If anyone is interested in doing a PhD or otherwise working with me on using NLG to improve personal health apps, please let me know!
PhilHumans will support 8 PhD students (paid as research assistants) who will look at using various types of AI (including NLG, NLP, computer vision, behaviour analytics, and machine learning) to enhance the way personal health apps communicate with their users. It is an industrial doctorate programme, which means that the students will spend at least half of their time working at a company (Philips), and indeed will look at using their ideas in the context of health apps sold by Philips.
To give an example of the vision (this is my example, its not an “official” example), Philips sells a number of sensors and apps that are intended to support new mothers, such as ugrow and baby+. In my vision, when a mother interacts with the app, the app will engage in a dialogue with her, which is informed by behaviour analysis (eg, when the baby usually sleeps), voice analysis (does the mother sounds stressed), computer vision analysis of the mother’s face (does she look stressed, tired, etc), as well as camera (using computer vision) and other sensors monitoring the baby. Based on this data, the app gives the mother information, advice, and/or reassurance (which is sometimes what new mothers most need!). The information is explained in an understandable way even for mothers who dont know a lot about health issues that affect infants and babies, and/or have limited numeracy or literacy.
In short, the app would interact with the mother in a way that hopefully is closer to the way in which an experienced nurse would interact with the mother. It would not make significant medical decision, but it would provide support, reassurance, and guidance, based on the state of the mother as well as of the baby.
We wont achieve this vision, but hopefully we will move closer to it!
As mentioned above, PhilHumans is an EU Industrial doctorate programme. This means that it funds PhD students who are paid salaries as research assistants (approx £30K/yr in UK) for three years. These are called Early Stage Researchers (ESR). They must
- have less than 4 years of research experience after their undergraduate degree (and not already have a PhD)
- must not have lived, worked, or studies in the host country (eg, UK) for the previous year (short visits and holidays are OK)
- be an EU national or otherwise able to work in the UK
There are 8 ESR positions in PhilHumans. The ones I am involved in are
- ESR7 (application page): This student will work on explaining relevant medical information to users with limited domain knowledge, numeracy, and/or literacy. It will involve NLG, and quite possibly dialogue and multimodal elements as well. This should be a very “hands-on” PhD, where the student does a lot of fieldwork with users and domain experts to understand the issues, designs and implements an algorithm, and then evaluates the algorithm with real users. The student will spend half of his/her time at Aberdeen, and half at Philips at Eindhoven. He/she will be supervised by myself and Steffen Pauws (Philips and Tilburg Uni). He/she will be based in Aberdeen, so cannot have lived, worked, or studied in the UK in the previous year.
- ESR1: This student will work on understanding the behaviour of users, using semantic and statistical modeling of the lifestyle and health status of the individual. The student will be primarily based at Philips Eindhoven, but will be registered for a PhD at Aberdeen, and spend six months at Aberdeen. He/she will be supervised by Aki Harma and Rim Helaoui (Philips) and myself. He/she will be based in Eindhoven, so cannot have lived, worked, or studied in the Netherlands in the previous year.
Students are expected to start in summer 2018.
These positions are not yet formally advertised, but please feel free to contact me informally if you are interested!