Some Thoughts on Building The Perfect Tool for Tracking Diet (Part 1)
A couple of weeks ago, I shared some lessons that I learned while tracking my diet for over six months. One of the conclusions was that tracking diet is one of the most cumbersome aspects of self-quantification/self-experimentation, mainly due to the lack of passive measurement tools and often overwhelming amount of nutritional information to collect and deal with. That naturally let me to the question: how can we make the diet tracking process easier? What would the perfect tool for tracking diet look like? The longer I am pondering this problem, the more I realize that we are talking not about a single stand-alone gadget or app, but rather a small ecosystem that includes digital scale, mobile phone and content. Let me now elaborate.
First, let’s look at what the ideal tool for tracking diet should be able to do:
In case of meals prepared at home:
1. passively or semi-passively identify the ingredients
2. weigh solid ingredients and measure volume of liquid ingredients, and automatically transmit that information to the central system
3. report and store nutritional information for the individual ingredients and complex meals
4. provide useful feedback and insights to the user based on historical data.
In case of meals eaten outside of home:
5. passively or semi-passively identify the ingredients or meal
6. passively or semi-passively collect, transmit and store nutritional information to the central system
7. provide useful feedback and insights to the user based on historical data
Now let’s try to “build” this tool. In today’s post, I will be discussing features 1-5 that apply to only cooked at home meals. We need to start somewhere, right?
The feature #1 is probably the most challenging one, from technological point of view. Even most advanced tracking gadgets and apps that exist at this point require some kind of manual action during this step. Digital nutritional scales, like Kitrics, require you to punch in the code of the ingredient using scale’s keypad. The Meal Snap requires you snap a picture of the ingredient or food item. The former approach is more inconvenient than the latter: you need to leaf through the booklet to find the code, and then use keypad to enter it. Of course, eventually you may memorize some of the codes, but that would take some time. Snapping a picture seems to be much easier, but the accuracy often suffers (although private parts of human body are correctly identified as inedible) The differences in nutritional values of similarly looking ingredients, like ham and turkey (or feta and cottage cheese) are significant, and photo analysis at this point can’t handle that differentiation. But I would not take the photo analysis off the table yet: extracting ingredient names from the photos of the grocery store receipts and restaurant menu descriptions could be potentially incorporated into the process. Image on the left: Kitrics digital nutrition scale allows you to track nutritional information of individual ingredients by entering code on keypad.
While the manual entry of ingredients can not be completely eliminated, it can be definitely simplified by means of well-designed mobile app. Imagine this. The app presents me with either the recipe of the meal (if this is the first time I am cooking) or the list of ingredients (if I cooked it before). Ideally, the app should handle both fresh ingredients (vegetables, meats) and processed foods (canned tomatoes, bread crumbs for frying fish). As I follow the recipe or go down the list, I place each ingredient on the scale, and check it off the list in my app. This is where the feature #2 comes in. The scale and app are connected wirelessly (Bluetooth or Wi-Fi); weight of each ingredient is transmitted back to the app, and the nutritional information is calculated. Think Withings for food. Even the volume can be derived through the weight; some digital scales like Taylor digital measuring cup/scale already do that. Some websites and apps try to eliminate weighing completely, by using instead ingredient sizes (e.g., “two medium potatoes”), but in our age, when strawberries can be bigger than potatoes, the size may not be an ideal reference metrics. Image on the right: Taylor digital measuring cup/scale allows you to convert weight to volume with one push of the button.
After I am done cooking, I save the meal information on the app. When time comes for lunch or dinner, I simply choose the name of the meal in the app, weigh individual portion on the scale and- voila! -nutritional information is calculated within seconds.
The key here would be content: library of simple recipes with no more than 6-7 ingredients, that are preferably grouped by dietary (e.g., Paleo, Atkins, low-fat, etc.) and lifestyle (15 minutes meals, raw foods, vegan, etc.) preferences. Think MyFitnessPal/FitDay/SparkPeople + AllRecipes, in one app, linked wirelessly to the scale. The library, of course, could include both pre-selected and user-generated content. I see a good potential for monetization here, too: instead of purchasing print cookbooks or diet guides, users could now purchase access to recipes, and even sell their own. And, of course, recipes should be completely customizable: for example, when following the recipe for chili, I should be able to replace ground beef with ground turkey with just a couple of swipes or taps.
I would also revise the nutritional information provided by this tool, giving users more options. For instance, some people, instead of exact amounts of macronutrients like fat, carbs, etc. may opt for just a total number of calories and whether their daily intake meets recommended dietary requirements. In other words, I do not need to know how much fat or sugars I consume, just need a nudge or warning when I exceed the allowed daily quota. Users can also opt for other metrics, like glycemic index, NutriPoints, ANDI, Weight Watchers points (Paleo Diet Points system, anyone?), or even build their own customized targets.
Finally, the tool can further engage users by providing interesting insights and feedback based on the historical nutritional and other data. For example, the tool can track accumulation of inflammation, cholesterol or sodium in the past several weeks, and recommend changes in the diet by offering specific recipes. For instance, if I have been eating too many inflammatory foods during last two weeks, the tool could warn me and offer me recipes with more anti-inflammatory ingredients. The library could also recommend recipes based on the time of the year (hot meals during fall time, cold meal during the summer; seasonal ingredients to save money, etc.), budget constraints, lifestyle goals (losing weight, preparing for marathon, gaining muscle, etc.). The possibilities are endless.
That’s it for today. I am still working on features #5-7 (tracking meals outside of home), and will be posting my thoughts when I am ready. Stay tuned!