Agenda ICT assisted dietary data acquisition an overview of novel technologies Mikkelsen, BE, Dobroczynski, M & Ofei, KA FENS Berlin, October 21, 2015 Abstract: Data collection in dietary intake studies using traditional methods is costly and time consuming and as a result the interest in ICT assisted automated or semi automated systems is high. Therefor there is considerable interest in methods that can assist this process. Tablet, smartphones in combination with imaging and vision technologies are some of solutions that have shown promising results. This presentation takes the Dietary Intake Monitoring System (DIMS) as a point of departure. It is a device for capturing accurate data on dietary intake. It has been developed for capturing information about patient s meal both before and after consumption in a foodservice setting and is used for assessment of food intake and plate waste. The DIMS is able to estimate the type and amount of food on a plate using an integrated technology based on imaging, weighing scale, IR thermometer and ID technology. The DIMS is used in a sequential mode: first the plated meal is recorded and second the returned plate is recorded. The 2 recordings then return the intake and the plate waste. The results so far indicates a substantial potential for decreasing the workload for registering food intake data. The paper discusses the DIMS technology, present other recent innovations such as the ebutton technology developed in the US and other recent ICT assisted approaches to measuring behaviour and suggest directions for future research directions as well as for research infrastructures in this field. Recent innovations NANA ebutton Foodscapelab & DIMS The future Technological challenges SenseCam & Meal Snap ebutton dvices for food project MTA Pittsburgh/Aalborg So, how does it work? While the company chalks it up to magic, we re assuming they ve got a handful of people (be it through Amazon s Mechanical Turk, or a room full of dudes promised free Internet in exchange for calorie counting) breaking down the meal in your picture item by item. Snap a shot of a chicken salad? They punch in some chicken, some lettuce, maybe some dressing and bam, they ve got a rough estimate. Is it a precise science? Hardly. Even in the screenshot above, you can see that there are some pretty wild variations. A Small handful of cashews, for example, comes back as being anywhere from 150 to 614 calories. Still, having some idea of what you re taking in is still far better than not having any idea at all. You can find MealSnap on the App Store for $2.99 right here [itunes link]. Gemming, et al., Eur J Clin Nutr (2013) 67, 1095 1099: Arab et al., Eur J Clin Nutr, 2011, 65(10):1156 1162, Video camera: looking at food and PA GPS: positioning the individual 3 axis accelerometer estimating motio 3 axis gyroscope: measuring body orientation UV sensor distinguishing indoor/outdo Barometer: determining body position/f level 1
Some ebutton issues Battery life (8h 14h) Sensitive to height (137 157 cm) Analyser time expenditure: 1 day footage = 1 day analysis Interclass correlation:.978.983 (visual).520.887 (ebutton) measured on calories Baranowski, T: Using the ebutton for measuring. dietary intake in children: Two formative studies, Training Course on ICT Assisted Methods for Measuring Diet & Behaviour in Complex Foodscapes, Aalborg University, 24 August 2015 Other systems ASA24 Automated Self Administered 24hdr TADA: Technical Assisted Dietary Assessment Diet Data Recorder System (DDRS) Smart Plate Smart Fork TelSpec http://www.tadaproject.org/ New technology in dietary assessment: A review of digital methods in improving food record accuracy, Stumbo, P, Proceedings of The Nutrition Society 02/2013; 72(1):70 6 Why are we doing this? Very First Prototype made by students from Integrated Food Studies i 2
Intelligent Buffet From DIMS1.0 to 1.5 Ofei, K. T., Holst, M., Rasmussen, H. H. & Mikkelsen, BE Effect of meal portion size choice on plate waste generation among patients with different nutritional status An investigation using Dietary Intake Monitoring System (DIMS). Appetite, 2015 DIMS 2.0 Inviting for co creation 3
Pre serve Post serve Intake Waste Input Output mode mode Weight Energy Protein Carbon Price eqv grammes kcal Grammes Grammes DKK DIMS ver. 2.5 Is the DIMS feasible? It reduces the time spent on NM from 15 to 4 minuttes Patients at nutritional risk produced increased amounts of plate waste, with less energy & protein intake when compared to patients not at nutritional risk. It can be used in co creation mode improving accuracy Ofei, K. T., Holst, M., Rasmussen, H. H. & Mikkelsen, BE Effect of meal portion size choice on plate waste generation among patients with different nutritional status An investigation using Dietary Intake Monitoring System (DIMS). Appetite, 2015 Conclusion Still a long way to go: algorithms and imaging Cross disciplinarity needed Privacy issues needs to be dealt with Distinguish between Pre Set and AllUImagine Non invasive computing opportunities: ICT assisted FFM s: pre post, type ID and amount ID Invasive computing cyborgs: Biosignals like throat sensor recording assessing portion/tooth implanted sensors that identify foods during consumption Thanks for your attention Graduate & PhD Course Measuring Dietary Behaviour the intelligent way Fudan University, Shanghai November 23 25, 2015 More info at www.dvices4food.aau.dk 4
Credits Dennis Godtfredsen Philip Brisson Martin Rene Andersen Patrick Lehmann Hald Stefanie Serafin Lars Reng Michal Dobroczynski Kwabena T Ofei Bent Egberg Mikkelsen Sanne Sansolios 5