Optifood – Feeding Vulnerable Children

Mathematically the problem is one of finding an optimum solution based on a variety of parameters and constraints, a situation that is exactly the approach of a technique called linear programming (go and read the Wikipedia article and then come and explain it to me if you understand!)

I felt pretty self-conscious as I sat down to lunch at WHO.  I was dining with about 20 academic nutritionists and I’d just realized that they had all chosen salads and water whilst I tucked into a particularly fatty steak, chips and bottle of fizzy orange (followed by a particularly obscene bar of chocolate).  I was suffering this professional scrutiny of my lamentable eating habits for the Optifood project, a WHO initiative in which b-i is working in close collaboration with the London School of Hygiene and Tropical Medicine and the USAID-funded Food and Nutrition Technical Assistance II (FANTA-2) project to create a diet analysis tool.

The project has pretty lofty aims – it hopes to revolutionize the way that nutritionists formulate recommendations to place their work on a sound mathematical footing.  It came as quite a surprise to me to learn that professional nutritionists have very little in the way of existing tools to help formulate food-based dietary recommendations.  Just how do they formulate food-based recommendations for a particular group of children who are deficient in iron and once identified how do they know that the iron-rich foods they recommend won’t leave the children deficient in another key nutrient?  Such analysis can be done but it is often very lengthy and time-consuming – finding just the right combination of foods to provide the required amount of up to 20 key nutrients.  It’s akin to solving several dozen equations at the same time and finding the best compromise between each.

Creating a formal tool presents quite a challenge as such a tool would need to take into consideration the local availability of foods, people’s preferences for what they eat and even social factors like taboo foods. In addition the nutrient requirements of a group varies by age and sex  and how much exercise they do so each analysis (to be truly useful) would need to be tailor-made with these factors considered.

Mathematically the problem is one of finding an optimum solution based on a variety of parameters and constraints, a situation that is exactly the approach of a technique called linear programming (go and read the Wikipedia article and then come and explain it to me if you understand!)

This is where the London School of Hygiene and Tropical Medicine came in – a senior academic there had created linear programming modules for nutrition using the MATLAB suite of software and needed a way of making these models available to a wider audience in a format that was much easier to use. USAID was interested in funding the development of the user-friendly format and support it with additional technical assistance from FANTA-2.

b-i was hired to address this problem, and our recommendation was to create a Windows Forms application for the math models to allow nutritionists to enter their data in an intuitive “wizard” interface and then to handle the data returned from MATLAB to create graphs, tables and comparisons.  The result is a tool that will be made available for free download from the WHO website and can be installed on any Windows PC to immediately begin analyzing diets and testing how possible recommendations will affect nutrient intake, which in turn affects health and, critical for babies and young children, early physical and mental development.

The tool is currently undergoing a year of pilot user testing with a number of sites around the world and some of the benefits are already being seen.  It’s creating quite a buzz; being able to provide hard mathematical data on diets is immensely useful for vulnerable groups of people but has wider implications in informing Government policy makers on feeding recommendations and replying to food industry groups who might want to lobby that their own (expensive) processed food should be given to young children because it is better than locally-sourced (cheaper) produce. Until now, the academic community had no quick or easy way to validate such recommendations from these industry groups who might not always be disinterested.

The tool might have implications for education, it would allow nutrition students to very quickly see the implications of changes to diet and there was talk of encouraging universities to include it in their Masters programs so that it could become a truly standard tool in the nutritionist’s tool box of tomorrow.

The tool could also be adapted to study toxins, and finding optimum diets that minimize exposure to naturally-occurring and man-made poisonous substances.

Finally, since the tool can be extended by the user to introduce new foods and new groups of people, it has applications to any situation and population – from refugees, those suffering famine, obese computer programmers or anyone with a medical condition that necessitates a particular diet (diabetics, allergies etc.)

Towards the end of the three day workshop the mood was positive and enthusiastic.  One professional is spear-heading a project in Cambodia to catch, wash, dry and cut up local tarantulas as a diet supplement (apparently they are an excellent source of protein and micronutrients, just watch out for the bite) and Optifood was going to allow her to see exactly how the arachnids would benefit her test group.  Another academic wanted to design a cheap diet supplement for sub-Saharan Africans and could use Optifood to see exactly which nutrients to target to help him find and justify funding.

With such lofty aims in mind therefore, I am quite happy to sacrifice my own waist line when I fail to resist the fried food on offer in the WHO canteen – although I certainly don’t propose running my own diet through Optifood anytime soon, although doing so might be a good way to test arithmetic overflow errors for the fat, salt and sugar variables.


  • Philippe Bogy
  • André Rios
  • Jonathan Bonvallat
  • Kevin Crampton

Digital news!

Are you as digitally addicted as we are? We can supply you with a regular dose of digital news. Simply sign-up, or click to Follow us online.

About the author

Kevin Crampton
Kevin Crampton