When you, like me, are on a 6-7 meals a day diet plan, the main challenge is to make sure that you are not getting hungry too soon between the meals, but feel hungry enough when the time comes for your next meal. That means carefully controlling the portions and ingredients that go into your food. Could quantification and self-tracking help in this case? This month, I started experimenting with “satiety” of food and ways to measure and predict it. In this post, I will discuss the properties of foods that are predictive of satiety and hungriness, including one particularly interesting diet metric: “fullness factor”.

For the past several weeks, I have been logging the following information about my meals (I also use 80Bites app to track the number of bites and time spent consuming the meal, but I will talk about that some other day):

- how hungry I am before the meal (hungriness)
- how “full” I feel after eating it (satiety)
- how tasty it was (gratification)
- weight of the meal (in grams)
- total number of calories
- and its fullness factor***

What is *fullness factor*? Have you noticed how some foods tend to fill you faster and prolong the feeling of “fullness” in your stomach? The numerous “satiety” studies have found that not only weight, but also some nutritional properties of food, like higher protein content, may increase meal’s satiety. The folks at NutritionData actually developed a mathematical model that estimates the satiating effect of the meal by looking at the nutritional properties of the ingredients. According to their analysis, satiety is directly related to nutritional density of the meal, specifically, total amount of calories, protein, dietary fiber and fat per 100 grams (you can see the actual formula on their website) . Using this model, NutritionData lab has derived a unique food metric that estimates how satiated the meal can make you, even before you consume it. It is measured on a scale from 0 to 5, and NutritionData database has fullness factors for most of the foods (including processed). For example, oranges have a slightly higher fullness factor (FF=3.5) than apples (3.3), so they are slightly more “filling”. So theoretically, by including the ingredients with higher FF, you could make your meal more “filling” without increasing its caloric content. I decided to test it using my own self-tracking data. The following analysis is based on three weeks worth of data that I collected in January. To track weight and nutritional content of my meals in this experiment, I used my new Kitrics digital nutrition scale.

First, I looked at how the meal’s satiety was correlated with its weight and total caloric content. Not surprisingly, there was a clear although modest relationship (Spearman’s correlation rho = .44) between these two: the heavier is the meal, the more “full” I felt after consuming it.

The same was true for total amount of calories in the meal, and the relationship was even stronger (Spearman’s rho = .58), and positive. The meals with more calories tend to be more “filling”.

I then looked at the fullness factor. By definition, for two meals with the same amount of calories, the one with the higher FF should be more “filling”. So I grouped meals by caloric content and weight, and then compared the average satiety ratings by fullness factor. Alas, I did not have enough data to get all possible breaks of weight x calories. Still, the data that ended up in the graph shows that my satiety ratings are clearly related to the fullness factor: even after controlling for calories and weight of the meal, the meals with higher FF had higher satiety score:

Another way to look at it is to compute my own satiety metrics. I simply divided my subjective satiety score by its total calories, and then plotted against the FF. There was a clear positive relationship between my own satiety score and NutritionData’s fullness factor (Spearman’s correlation rho = .52):

In summary, my own self-tracking data helped me to demonstrate that fullness factor can be used in planning meals. Including ingredients with higher fullness factor can help me to keep my meals hearty and satisfactory without increasing their weight or caloric contents. Stay tuned for more Quantified Self diet hacks!

***Computing the fullness factor for composite meals is a bit tricky. I tried using NutritionData’s *Create Recipe* feature that automatically calculates the fullness factor, but it did not work (I tried both Chrome and Internet Explorer). I tweeted to NutritionData, but to no avail. So I decided to calculate it myself. I assumed that the composite fullness factor is proportional to ingredient’s share in the recipe. Let’s take, for example, panini. I have had it for 5 days for either lunch or dinner. The ingredients were whole wheat roll (150 grams; FF = 2.1), smoked turkey (100 grams; FF = 3.1), and avocado (50 grams, FF = 2.3). The total weight of the meal is 300 grams, and the ingredients’ shares are 50% (whole wheat roll), 33.3% (turkey), and 16.6% (avocado). Then the Fullness Factor of panini: FF(panini) = 50% of FF(panini) + 33% of FF(turkey) + 16.6% of FF(avocado) = 0.5*2.1 + 0.33*3.1 + 0.17*2.3 = 1.05 + 1.02 + 0.39 ~2.46, which I would round up to 2.5. Using the same principle, I went through my food log for the past 3 weeks, and computed fullness factor for each meal.