Delivering customized nutrition with cognitive insights
Nutrition is the science of how food effects the human body and focuses upon disease prevention, healing and management of chronic conditions. A dietitians’ field of work is however much generalized. This includes working with different diets, applications, data sources, articles, and multiple recommendations, that sometimes contradict each other. Yesterday we were told that eating eggs is not good for our health, and cereals in the morning for our kids is the best choice, today we see a different picture. You may ask what it is to do with information, and particularly governance and analytics?
Well, that is exactly the case. Food industry, and nutrition specifically, is one of the most complex sciences. As if we look at the various stakeholders and players, we see that they come from different industries, have different use cases, but at the end they all touch nutrition and overlap with each other. Nutrition assessment takes many forms and is driven by many companies. It is covered by insurance providers, computerized food systems at the hospitals, food supply chain and distribution that affect food safety, retail and online shopping for food grocery, healthcare providers who look at nutrition as one of the preventions for NCDs (Non-Communicable Diseases), medical devices companies like Medtronic who provide applications for diabetics, and the list goes on.
And here is another challenge that we see – what are the data sources to use, what are the attributes, and are there any standards? When, what, how, why, where did I eat? What was the weather? My preferences? How does my metabolism work? Are there any genetic factors that we need to account for that may contribute to obesity or other disorder? What are the cultural norms that may affect eating habits? Lifestyle? Physiological, habits, medical history, nutrition deficiency, age, race, sex, weight, height, etc. All this seems like a big 3D jigsaw puzzle.
We can see the complexity of nutrition science and nutrition information, and this is where analytics and cognitive approach can help. Today we use limited data sources, and recorded personal data is not always aggregated and compared with other similar use cases across population. Lack of integration between clinical, nutritional, behavioral/physical activity, genomic data is one of the limitations. What if we can understand and collect all the data that we need. What if we can create special nutrition models that can be used for estimating human requirements and
- nutrients derived from menus created for specific disease, and
- used by population based on geography (location, population density), demographic (age, race), physiological (personality, social class), education (knowledge about medical condition),
- medical pre-condition (genetics, family history, personal habits – smoking, drinking, etc.)
Insights from such data can be used to enhance or modify a person’s diet thereby improving and protecting their health. Such insights can be priceless, because we are moving to personalized aspect of well-being.
The lack of information from the food industry, combined with the complexity of nutrition sciences research – creates another significant challenge. This challenge can again, be addressed using the power of cognitive computing, machine learning and data science. With a reduction in complexity and proper user experience testing and design, changing one’s diet can become an enjoyable experience.
Learning and building knowledge from various structured and unstructured sources of information, understanding natural language and interacting more naturally with humans, help us to build new knowledge systems. Learn more about Bluemix Data Connect.
And with knowledge we can reduce the complexity and present an understanding about good food, GMF (genetically modified food), specific practices, awareness and motivation. Knowledge will be the key to those community and medical leaders who wish to influence food choices for the better. By collecting the data from the patients, such as their feedback on the food/recipe selected, feeling condition, etc., de-identifying it with encoded nutritional relationships with medical and physiological effects related to the patient, and by applying cognitive models that we will be able to train as the data is collected, we will be able to provide insights to understand the specific impact of the nutrition food on the disease. But when again, we should not forget that this is very sensitive data, and we have even more challenges than we think, when we assume it is very simple subject.
In this complex environment where information is passed in and out from multiple mobile devices to cloud for all other information collected from other consumers for all data transactions, tracking locations, personal eating habits, and other personal data- we start to store increasingly more and more personal or otherwise sensitive data. And if we are bringing here healthcare personal data, we face even bigger challenge.
There are regulations in many countries that govern the privacy and security of healthcare data. HIPAA in USA, POPI in South Africa are just few examples, that enforce certain obligations on anyone who stores or processes healthcare data.
Here it is important to have a platform that provides the capabilities for secure data protection and governance. The sensitivity of data and its criticality are important factors that we need to keep in mind when we are dealing with such simple on the first glance thing – as food and nutrition. We need to develop and clearly specify secured access policy, taking into consideration partnership across geographical borders. We need to define and enforce governance rules, properly classify the data. Models of common subject areas, such as consumer, food order, recipe, could provide a starting point for the subject area teams. There could be on premises data sources or cloud data sources, external data such as social, weather data – all in different format, including video. Data still need to be transformed, cleansed and prepared using data integration technologies. We need to store the data, and drive the analytics with models that will evolve over time as we collect more and more data, that allow us to deliver the right information in context to users based on their personalized profile and needs
In summary, now we had collected nutrition and health data, built up the correlation between them. With computational method and analysis, we know how all these factors may affect certain medication situation. We can improve user experience with social aspect, delivering positive message, reducing the complexity of nutritional sciences into highly personalized and digestible amounts of information.
I love food. And what we eat matters, and it can really change our lives. And I also want to be sure that the system I use for receiving these customized insights for me, can integrate multiple sources of data in secure and protective environment, understanding the correlation and evolving over time. Today such a system exists, and it is called the Watson Data Platform.