In this digital age, data is generated monstrous from diverse sources like IoT enabled smart gadgets, and so on worldwide very swiftly in distinctive formats. This data with the traits say volume, velocity, variety and so on referred to as big data. Since a decade, big data technologies have been utilized in most of the companies including healthcare sector.
Healthcare sector is not only using big data but also using IoT to gain treasured insights in making knowledgeable decisions spontaneously to improve medical treatment particularly for patients with complicated medical history having multiple health ailments. For healthy living, after water and oxygen, diet plays a critical role in offering the strength needed to assist the life’s existence-maintaining strategies and metabolism.
The intent of this work is to offer a framework that classifies the population into four classes based on the quality of diet they devour within 30-days of dietary recall as balanced, unbalanced, nearly balanced, and nearly unbalanced using the machine learning techniques specifically logistic regression, linear discriminant analysis(LDA), and random forest. NHANES datasets had been used to assess the proposed framework alongside the metrics accuracy, precision, etc.
This framework also allows us in gathering person’s health and dietary details dynamically anytime with the voice (IoT) to find out to which food regimen the person belongs to. This could be beneficial for medical doctors, dieticians, and also to an individual.Keywords — Healthcare, Machine Learning, IoT, Nutrition, Big Data.
Big DataBig data can’t be affixed with categorical source as its miles an explosion of data. This explosion is recursive and illimitable; its miles perpetually evolving and dynamic. This has engendered a buzz about the challenges gigantic information offers. Big data are created from monstrous amounts of facts of a ramification of media types (photos, audio, video, textual content, parameter measurements etc.) and shape (structured, semi-structured, and unstructured) unpredictably coming in near real-time from multifarious sources (namely, traditional, web-server logs, and click-stream data, social media reviews, phone call records, wearable data, RFID tags, smart gadgets and data captured via sensors through IoT kits) to be related, matched, cleansed, and converted across systems. Big data is not simplest approximately its size nevertheless concerning the value within it.
The 5 Characteristics of Big Data (adopted from Haas, 2013)
Over the last decade, big data frameworks like Apache Hadoop alongside its ecosystem components like Apache Pig, Apache Hive, Apache Flume, Apache Sqoop, Apache Mahout, and many others have been utilized in most of the organizations, including healthcare, to extract valuable insights from this commodious, multifarious data (patient health records, lab reports, treatment data, and many others) to carry out their operations efficaciously, efficiently, and in a cost-effective manner. Big data analytics is facilitating healthcare vicinity to store and make informed choices spontaneously to improve the affected person’s treatment, especially for patients with complicated medical histories, tormented by more than one complaint.
Healthcare and IOT
The sedentary nature of work and modern food habits may cause long-lasting illnesses which include cardiovascular ailment , hypertension, stroke, diabetes, overweight, and many others. The increased cost of healthcare offerings has expedited the stress among the sufferers and additionally to the regimes in getting or offering potent and efficient healthcare in many of the developing nations. As a complex cyber-physical system, IoT amalgamates all kinds of sensing, identity, communication, networking, information management devices and systems, and seamlessly links all of the human beings and things consistent with the pastimes, in order that anyone, at any time, and everywhere, through any tool and media, can get access to any data of an object to achieve any service more efficaciously(ITU 2005; European Commission Information Society 2008, 2009).
The effect as a result of the IoT to human society will be as big as that the world wide web has prompted long time back, so the IoT is acknowledged as the ‘subsequent generation of internet’. IoT equation can be formed as: IoT = internet + physical objects + controllers, sensors, and actuators.
IoT permits gadgets discerned or administered remotely across existing network infrastructure, developing opportunities for the greater direct amalgamation of the phenomenon into PC-based systems, and resulting in improved efficiency, accuracy and economic advantage in addition to reducing human involvement.
Balanced Vs Unbalanced (Malnutrition)
DietIn today’s lifestyle, Malnutrition accounts to be a huge hassle. Malnutrition is a condition as a result of consuming meals wherein nutrients are either not enough or too much such that it causes health problems. It could involve protein, carbohydrates, nutrients, or minerals. Not enough vitamins are called under-nutrition and the reverse of it is referred to as over-nutrition. Malnutrition is typically used in particular to confer with under-nourished where a man or woman constantly gets inadequate strength.
The Balanced diet is that diet, which is rich in nutrients. It includes whole grains, fruits, vegetables, dairy products, etc., when taken supplies proteins, carbohydrates, vitamins, minerals, fiber, and fat, etc., needed to help maintain individuals health and to protect from diseases. However, unbalanced diet is food regimen, which consists of either fewer or extra of the nutrients than your body wishes. Moreover, nutrient imbalance leads to deficiencies, obesity(weight gain) and also affects the immune system of a person adversely. Recent arena of disease study reveals that the poor diet is one of the main factors in one among the five deaths worldwide.
Moreover, as per World Health Organization (also called WHO) and other sources, there is nearly a tenfold increase of obesity in children, adolescents, and adults for the past four decades by continuing the same trend, it is expected that the world will have more obese people than no obese people by 2030 thereby leading to non-communicable diseases (NCDs) like hypertension, kidney problems, diabetes, heart diseases, cancer, etc. Consumption of unhealthy diet is causing non-communicable diseases (NCDs) and other health ailments.
According to the WHO’s report, approximately 2.7 million deaths are happening due to NCDs each year. To reduce the no. of deaths, WHO released the guidelines to the health care workers to actively identify and manage, especially children who are obese. The goal here is to pick out parameters that categorize dietary intake quality ate up by the person into balanced, nearly balanced, nearly unbalanced, unbalanced food regimen and also explanatory elements which have an effect on those nutrition defining guidelines.
The principle contributions of this paper
On this paper, we suggest a Population Classification Using Dietary Data (PCUDD) framework for enhancing the working efficiency and reducing the operating time of nutritionists, individuals, and medical doctors in determining the kind of the diet taken by a person and their associated risk factors. In this paper, classification results were given on the NHANES datasets that are cleaned and pre-processed, and compare the results of multinomial logistic regression, LDA, and random forest algorithms.
PCUDD’s performance can be tested by real datasets extracted from any individual with dietary recall information.As per our experimental results, the PCUDD can attain a mean accuracy of 87% for classifying populace diet as a representative example. The outcomes imply that the PCUDD can assist medical doctors/dieticians with the aid of speedy narrowing the scope of diagnosis, thereby satisfying the objective of increasing the performance and decreasing their work burden.