Published on 30.01.04 in Vol 6, No 1 (2004)
DietPal: A Web-Based Dietary Menu-Generating and Management System
Background: Attempts in current health care practice to make health care more accessible, effective, and efficient through the use of information technology could include implementation of computer-based dietary menu generation. While several of such systems already exist, their focus is mainly to assist healthy individuals calculate their calorie intake and to help monitor the selection of menus based upon a prespecified calorie value. Although these prove to be helpful in some ways, they are not suitable for monitoring, planning, and managing patients' dietary needs and requirements. This paper presents a Web-based application that simulates the process of menu suggestions according to a standard practice employed by dietitians.
Objective: To model the workflow of dietitians and to develop, based on this workflow, a Web-based system for dietary menu generation and management. The system is aimed to be used by dietitians or by medical professionals of health centers in rural areas where there are no designated qualified dietitians.
Methods: First, a user-needs study was conducted among dietitians in Malaysia. The first survey of 93 dietitians (with 52 responding) was an assessment of information needed for dietary management and evaluation of compliance towards a dietary regime. The second study consisted of ethnographic observation and semi-structured interviews with 14 dietitians in order to identify the workflow of a menu-suggestion process. We subsequently designed and developed a Web-based dietary menu generation and management system called DietPal. DietPal has the capability of automatically calculating the nutrient and calorie intake of each patient based on the dietary recall as well as generating suitable diet and menu plans according to the calorie and nutrient requirement of the patient, calculated from anthropometric measurements. The system also allows reusing stored or predefined menus for other patients with similar health and nutrient requirements.
Results: We modeled the workflow of menu-suggestion activity currently adhered to by dietitians in Malaysia. Based on this workflow, a Web-based system was developed. Initial post evaluation among 10 dietitians indicates that they are comfortable with the organization of the modules and information.
Conclusions: The system has the potential of enhancing the quality of services with the provision of standard and healthy menu plans and at the same time increasing outreach, particularly to rural areas. With its potential capability of optimizing the time spent by dietitians to plan suitable menus, more quality time could be spent delivering nutrition education to the patients.
J Med Internet Res 2004;6(1):e4
- Dietary services;
- menu planning;
- information systems;
- health information system;
- Web-based services;
Planning nutritious and appetizing menus is a complex task that researchers have tried to computerize since the early 1960s [, ]. Although a number of menu-planning systems have been developed in recent years, these systems are mainly used to assist healthy individuals calculate their calorie intake and to help monitor the selection of menus based upon a prespecified calorie value. Also, some of these systems do not address the standard practice and procedure employed by dietitians during consultations with patients. Planning nutritious menus for patients, however, is not the same as planning menus for healthy individuals. Patients require advice and directions from dietitians in designing their menus. Dietitians on the other hand, during the course of consultation with a patient, may want to refer, for example, to the patient's medical and dietary history, dietary recall, biochemical data, and anthropometric data. to construct a suitable dietary plan and menu for patients. For example, a suitable menu for a diabetic patient is constructed based on the patient's calorie requirement determined using anthropometric data such as weight and height. The patient's blood sugar control is examined to decide on refined-sugar allowances. Finally, obtaining the dietary history of a particular patient is necessary in order to consider factors such as food habit and preferences. Therefore, an ideal menu-planning system should not only contain information about foods and menus but should also incorporate other related information for the purpose of decision making by dietitians (as discussed earlier).
Realizing the limited capabilities of existing systems, this paper, therefore, describes the development of an automated Web-based menu-generating system, according to a standard procedure and practice adhered to by dietitians in managing patients, and based upon a user-needs study conducted prior to the development of the system. This project is a collaboration between the Faculty of Information Science and Technology and the Faculty of Allied Health Sciences (see "Acknowledgements" for funding information).
At the moment, the system is intended for use by dietitians and health professionals to extract patients' dietary recalls and to design suitable menus based on a patient's dietary habits and nutritional requirements.
Related Research and Development
As previously mentioned, there have been a number of menu-generating systems or dietary-analysis programs available, either implemented as a Web-based application [- ] or as a traditional information system [ - ].
A few systems reviewed—such as Case-Based Menu Planner (CAMP) , Pattern Regulator for the Intelligent Selection of Menus (PRISM) [ ], and CAMP Enhanced by Rules (CAMPER) [ ]—employed techniques from the field of artificial intelligence (AI):
- CAMP employs the case-based reasoning (CBR) technique to suggest menus to users. CAMP uses past menus that were compiled from reputable sources and modified as needed to ensure that they satisfy the RDIs (Reference Daily Intakes) and the Dietary Guidelines of Americans and Aesthetic standards [ ]. The menu generated by CAMP is based upon nutrient composition, type of servings, and the number of snacks.
- PRISM uses rules to generate menus. The rules are mainly concerned with menus and meal patterns.
- CAMPER is an integration of the techniques employed by CAMP and PRISM. Therefore, apart from using the case-based reasoning technique, CAMPER uses rules or "what if" analysis module to enhance the menu suggestion activity.
The main interesting and distinguishing feature exhibited by our system as compared to the other reviewed systems is its use of the complete dietary-management system currently adhered to by dietitians in Malaysia, particularly at the National University Hospital of Malaysia. In addition, our system exploits current advanced Internet technology, by considering the system's implementation as a Web-based application. This, to a certain extent, increases the outreach of the system for use by dietitians and health professionals within the same hospital or at other locations. The system is also capable of storing and organizing patients' dietary records and other health-diet related information. This capability would allow dietitians to effectively evaluate or monitor the patients' dietary changes throughout the period of consultations.
User Information Needs Study and Functional Specification
Prior to the development of the system, a user-needs study was conducted among dietitians within the Klang Valley in Malaysia .. The user-needs study included 2 independent surveys. The first survey of dietitians was on computer literacy and utilization, information seeking activities, and assessment of information required for dietary management and evaluation of compliance towards a dietary regime. The second survey consisted of semistructured interviews with a subsample of dietitians in order to extract the knowledge and workflow of a menu-suggestion process.
Development of DietPal
The development of DietPal took into account the key findings of the user-needs study and the consultation flow currently adhered to by dietitians.
An initial post-evaluation has been conducted by distributing a written survey among 10 dietitians who are directly involved in the management of patients from the National University of Malaysia Hospital.
User Information Needs Study and Functional Specification
In the first survey, questionnaires were posted to 93 clinical dietitians registered with the Malaysia Dietitians' Association; 52 subjects (56%) responded. The questionnaire contained the question "Which information items do you seek to support decision making and evaluation of a dietary regime", with the 10 items listed inand participants asked to assign a rating score of 1 to 5 (1 = least likely seek, 5 = most likely seek) was used to identify the information needed to. In a similar way, participants were asked which information items they needed in order to monitor the compliance of a patient ( ).
and present part of the results obtained from this user-needs study. Because the questionnaire was self administered not all items were responded to by all 52 subjects; thus, missing data was unavoidable.
Based upon the results presented inand , the most common information used to support decision making according to the mean of scores were medical diagnosis, current body weight and height, nutrient requirement, biochemical values, and dietary recall.
The most relevant indicators for monitoring patient compliance were reported to be biochemical values, food intake changes, compliance to diet regime, and weight change.
The second survey involved ethnographic observation of 14 clinical dietitians from government, university hospitals, and private hospitals doing individual dietary counseling. This study was similar to an information-engineering activity, in which the task was to extract the knowledge and the workflow of generating a menu for patients. In this case, 14 dietitians were interviewed and observed while conducting their daily activities. The survey was also aimed at validating the results of the first survey and determining at which stage of the menu-suggestion process the "frequently sought after" information items are being used. The result of this survey is a workflow of a menu-suggestion process currently used by dietitians in Malaysia, as illustrated in.
Menu Generation and Management with DietPal
As illustrated in DietPal's system architecture (), the system consists of 5 databases with 2 main modules, the Management module and the Menu Generating Module. The Management module is designed to manage information relating to patients' personal and medical information and to assess patients' compliance to dietary regime. The Menu Generating module is designed to interactively assist users in planning suitable menus and diet plans for patients. The Patient database is used to store the patient's information, which includes the personal data, anthropometric and biochemical data, medical record, and information on dietary recall. This database is heavily used in the Management module. The Food Composition database consists of information about foods and nutrient composition, extracted from the Malaysian Food Composition Tables for macronutrients and micronutrients. This database is used for food-analysis purposes. The Diet Plan/Menu database consists of therapeutic menus for specific diseases; the menus were obtained from reputable sources and have been approved by dietitians.
When using DietPal for an existing patient, the dietitian is first required to update the anthropometric data, particularly the body weight; this is to record any important patient changes since the previous meeting. If current biochemical data is available, such data will also be recorded by the dietitian; this is to monitor the patient progress so as to evaluate patient compliance to the previously-suggested menus. Information regarding any new medical diagnosis and current medication, if any, will also be recorded. For new patients, the dietitian is required to formally register the patients with the system; information such as name, date of birth, address, occupation, and other information similar to that for existing patients will be recorded.
The next step involves acquiring the patient's dietary recall in order to assess the patient's compliance to a dietary regime as well as to assist dietitians in planning suitable menus based on the patient's food habits. In normal circumstances, the dietitian will record the patient dietary recall of up to 7 days of food intake. In this case, the dietitian will interview the patient and will select the available food stored in the food-composition database. For cases in which the food is not available in the database, the dietitian will choose other similar types of food based on calorie content. DietPal will automatically calculate the level of macronutrients and micronutrients for an average of 1 day. Apart from that, DietPal also allows the user to compare the level of macronutrients with the individual calorie requirement based on Basal Metabolic Rate (BMR), physical activity, and stress factor, while the micronutrient intake is compared with the Malaysian Recommended Dietary Allowance (RDA)  for nutrients determined according to the patient's sex and age group. The task of dietary recall is important for assessing the patient's compliance to the dietary regime, but, because this process is time consuming, it can be passed over by the dietitian, who can go directly to the menu-generating function.
The menu generation starts with DietPal automatically calculating the Body Mass Index (BMI) and providing a suggestion as to whether the patient is, for example, normal, obese, or underweight. Simple rules are used to make such a decision. Anthropometric information is also displayed by DietPal in order to assist dietitians in making decisions. Two methods can be used to automatically generate the energy requirements of each individual patient:
- Predictive equation to estimate energy requirement based on Basal Metabolic Rate (BMR) ( ), stress factor ( ), and activity factor ( ). The energy requirement is estimated using the following formula [ ]:Energy requirement = BMR × stress factor × activity factor.
- Quick method, based on the following formula:Energy requirement = weight (kg) × quick method factor (kcal/kg)
Values of the quick method factor based on weight status and physical activity are shown in[ ].
Once the energy requirement has been determined, the macronutrient requirement will be calculated (the energy is distributed into 3 macronutrients, see). In this case, the dietitian will provide the percentage of carbohydrate, protein, and fat (C-P-F) and DietPal will automatically calculate the kilocalories and grams of carbohydrate, protein, and fat according to the energy requirement of the particular patient. Values in are the default values provided by the system; they are derived from the Malaysian Dietary Guidelines [ ]. Dietitians, however, are allowed to alter these values based on patients' requirement as illustrated in . If the total percentage of all nutrients exceeds 100% or is below 100% an error message is provided to the user. Based on these nutrient percentages, the kilocalories and grams of the nutrient are automatically generated by the system.
The next step requires the dietitian to design the exchange table by food groups and by meals. An exchange table describes the portions of each food group based on a prefixed amount of carbohydrate, protein, and fat. For example, 1 exchange of food from the cereals food group contains 15 g of carbohydrate, 2.0 g of protein, and 0.5 g of fat. In this case, DietPal will first suggest an exchange table based on the standard calculation currently practiced by Malaysian dietitians. The user is allowed to alter any exchange portions according to the patient's needs; DietPal will update the amount of carbohydrate, protein, fat, and energy accordingly. Once the exchange table by food group has been successfully produced, the dietitian will continue with the design of an exchange table by meal. In this case the dietitian is required to fill in the exchange portions for the relevant mealtimes. DietPal will detect any inconsistency that might occur, such as if the distributed amount of exchange portions is not equal to the total exchanges.
As an alternative to the aforementioned process, the dietitian can retrieve the standard existing exchange table by food and by meal stored in the Diet Plan database. To date, DietPal has a number of diet plans ranging from 1200 kcal to 2000 kcal designed by dietitians from the Faculty of Allied Health Sciences of the National University Hospital of Malaysia.illustrates the output of the exchange table by food and by meal for the 1600 kcal standard diet plan. The distribution of the macronutrients are based on a prefixed range of exchanges of each food group (ie, cereal, 8-14 exchanges; fruits, 2 exchanges; skim milk, 1-2 exchanges; vegetables, minimum 2 servings; meat, 1-2 exchanges; fish, 2 exchanges; and oil, 6-10 exchanges). The use of existing diet plans greatly reduced the time for designing such food-distribution tables.
The menu suggestion will be made after all the aforementioned steps have been completed. There are currently 2 available approaches offered by DietPal—either dietitians design the menu based upon the nutrient requirement and exchange tables or they can retrieve and reuse preexisting menus from the Case Base. From the exchange allowances of each food group (), the dietitian will then be able to generate a suitable menu based on the Malaysian Food Composition Table. Alternatively, the dietitian can retrieve preexisting menus based upon the nutrient requirement and the disease state of the patient. The retrieved menu can be edited or altered by dietitians according to the suitability of the patient. The output menu—either manually-designed or retrieved from the Case Base—will be stored in the Case Base for future use. Dietitians can design or retrieve more than 1 menu for the patient. illustrates an example of a menu plan generated using DietPal.
Medical professionals of health centers in rural areas where there are no designated qualified dietitians will find the preexisting menus useful for advising patients with certain disease problems. The menu designed is only suitable for nonvegetarians, as vegetarianism is not prominent within the Malaysian society.
The functionalities of DietPal are distributed to a number of different levels of menus (pages) as illustrated in, which correspond to the menu-suggestion activities previously described.
A preliminary evaluation has been conducted among 10 dietitians who are directly involved in the management of patients from the National University of Malaysia Hospital and who used the system. The result of this evaluation is illustrated in, where scale = 1 represents strongly disagree/unsatisfied and scale = 5 represents strongly agree/satisfied. The categories used during evaluation were derived from [ - ].
Results fromshow that the system scores above average for all the categories evaluated, with satisfy user's requirement scoring the highest marks. These results indicate that on average dietitians are satisfied with the overall capability of the system to generate and manage dietary menus.
The adoption of Internet- or Web-based technology in health-related applications is still lagging well behind adoption in other fields . It is expected that if communications through the Internet and the World Wide Web are looked into seriously, the efficiency of delivering health care and services could be increased [ ].
DietPal is a dietary menu generation and management system for patients, with simple intelligent capabilities to design and generate suitable diet plans and menus based upon the patient's energy requirement. Comparing this system with other similar Web-based systems—such as the Menu Planner developed by the National Health, Lung and Blood Institute (NHLBI) ; the Nutrition Analysis Tool (NAT) [ ], a public service system provided by the Food Science and Human Nutrition Department at the University of Illinois; and DietSite.com [ ], provided by Dietsite.com Inc—the system exhibits a few distinguishing features:
- capacity to store and retrieve historical data related to a patient, which can be used to assess the patient's compliance to the dietary regime
- retrieval of preexisting menus that suit the requirements of an existing patient
- capacity to monitor the progress of each patient based upon the menu suggested
More rigorous testing and evaluation of the system is currently ongoing. Based on the preliminary experiences reported here, we think that it has the potential to provide the following significant contributions:
- Assist in enhancing the quality of health services and improving the outreach in urban and, particularly, rural areas with minimal costs.
- Provide standards for healthy menus for patients. Hospitals and clinics in rural areas in particular, therefore, will be able to access an up-to-date database specifying the needs of certain patients.
- Optimize time spent by dietitians to calculate nutrient intake and energy requirements, and to generate exchange tables and menus. Thus, dietitians will have more time to deliver nutrition education to patients. A cost-effectiveness study on the usage of the system among dietitians or health professionals will be conducted.
Our ongoing work includes applying artificial-intelligence techniques to intelligently generate suitable menus for patients with certain diseases such as diabetes, hyperlipidaemia, obesity, and hypertension. We are exploring the case-based reasoning technique, where a new problem is solved by finding similar past cases, and reusing them in the new problem situation . Although such a technique has been used in CAMP [ ] and CAMPER [ ], our approach differs from those two in the following aspects:
- We will consider the 4 major diseases (diabetes, hyperlipidaemia, obesity, and hypertension) in generating the required diet plans or menus.
- As Malaysia is a multiracial country (eg, Malay, Chinese, Indian), we need to consider menus according to the respective ethnicities.
- We will consider the patient's dietary recall as well as the patient's dietary exchanges in providing suitable menus.
We gratefully acknowledge the funding support we have received from the Malaysian Ministry of Science, Technology and the Environment (MOSTE) under the 7th Malaysian Plan of Intensification of Research in Priority Areas (IRPA), Project No: IRPA 04-02-02-0035. We thank all the dietitians involved in this study.
Conflicts of Interest
- Balintfy JL. Menu planning by computer. Communications of the ACM 1964;7(4):255-259. [CrossRef]
- Eckstein EF. Menu planning by computer: the random approach. J Am Diet Assoc 1967 Dec;51(6):529-533. [Medline]
- ; Center for Nutrition Policy and Promotion, US Department of Agriculture. Interactive Healthy Eating Index (IHEI). URL: http://www.usda.gov/cnpp/ihei.html [accessed 2003 Nov 4] [WebCite Cache]
- ; Fitday.com. Home page. URL: http://www.fitday.com/ [accessed 2003 Nov 4] [WebCite Cache]
- ; Cyberdiet. Home page. URL: http://www.cyberdiet.com/ [accessed 2003 Nov 4] [WebCite Cache]
- ; Heart, Lung, and Blood Institute; National Institutes of Health. Interactive menu planner. URL: http://hin.nhlbi.nih.gov/menuplanner/ [accessed 2002 Aug 12] [WebCite Cache]
- ; National Institutes of Health. Nutrition Analysis Tool 2.0. URL: http://nat.crgq.com/ [accessed 2002 Aug 14] [WebCite Cache]
- ; DietSite.com. Home page. URL: http://www.dietsite.com/ [accessed 2002 Sep 21] [WebCite Cache]
- Marling CR, Petot GJ, Sterling L. Integrating case-based and rule-based reasoning to meet multiple design constraints. Computational Intelligence 1999;5(3):308-332. [CrossRef]
- Petot GJ, Marling C, Sterling L. An artificial intelligence system for computer-assisted menu planning. J Am Diet Assoc 1998 Sep;98(9):1009-1014. [Medline] [CrossRef]
- Sterling L, Petot GJ, Marling CR, Kovacic K, Ernst G. The role of common sense knowledge in menu planning. Expert Systems with Applications 1996;11(3):301-308. [CrossRef]
- Marling CR. Integrating case-based and rule-based reasoning in knowledge-based systems development [dissertation]. Cleveland, Ohio: Case Western Reserve University; 1996.
- Kovacic KJ. Using common-sense knowledge for computer menu planning [dissertation]. Cleveland, Ohio: Case Western Reserve University; 1995.
- Rafidah G, Suzana S, Shahrul Azman MN, NoorAini MY. Identification of information needs for developing a computerised dietary management system among health professionals in Kuala Lumpur and Selangor. Presented at: 3rd Asian Congress of Dietetics; 2002; Kuala Lumpur.
- Teong TS. Recommended daily dietary intakes for peninsular Malaysia. Med J Malaysia 1975 Sep;30(1):38-42. [Medline]
- Schofield WN. Predicting basal metabolic rate, new standards and review of previous work. Hum Nutr Clin Nutr 1985;39 Suppl 1(Suppl 1):5-41. [Medline]
- Cole HS, Camerini-davalos RA. Diet therapy of diabetes mellitus. Med Clin North Am 1970 Nov;54(6):1577-1587. [Medline]
- ; National Coordinating Committee on Food and Nutrition. Malaysian dietary guidelines. Kuala Lumpur: NCCFN; 1999.
- Kumar K. Post implementation evaluation of computer-based information systems: current practices. Communications of the ACM 1990;33(2):203-212. [CrossRef]
- Saracevic T. Evaluation of evaluation in information retrieval. In: Sigir 95: 18th Annual International Acm-Sigir Conference on Research & Development in Information Retrieval. New York, NY: Association for Computing Machinery; Dec 1, 1995:138-146.
- Keevil B. Measuring the usability index of your web site. In: Proceedings of the 16th Annual International Conference on Computer Documentation. New York, NY: Association for Computing Machinery; Sep 24, 1998:271-277.
- Coiera E. The Internet's challenge to health care provision. BMJ 1996 Jan 6;312(7022):3-4 [FREE Full text] [Medline]
- ; Arthur D Little Inc. Telecommunications: can it help solve America's health care problems?. Cambridge, MA: Arthur D. Little; 1992.
- Aadmodt A, Plaza E. Case-based reasoning: foundational issues, methodological variations and system approaches. AI Communications 1994;7(1):39-59 [FREE Full text] [WebCite Cache]
|BMR: Basal Metabolic Rate|
Edited by G. Eysenbach; submitted 06.03.03; peer-reviewed by P Stumbo, R Weiss; comments to author 24.06.03; revised version received 24.08.03; accepted 17.10.04; published 30.01.04
© Shahrul A Noah, Siti Norulhuda Abdullah, Suzana Shahar, Helmi Abdul-Hamid, Nurkahirizan Khairudin, Mohamed Yusoff, Rafidah Ghazali, Nooraini Mohd-Yusoff, Nik Shanita Shafii, Zaharah Abdul-Manaf. Originally published in the Journal of Medical Internet Research (http://www.jmir.org), 30.1.2004. Except where otherwise noted, articles published in the Journal of Medical Internet Research are distributed under the terms of the Creative Commons Attribution License (http://www.creativecommons.org/licenses/by/2.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited, including full bibliographic details and the URL (see "please cite as" above), and this statement is included.