[Previous Article] [Next Article] [Português] [Español]

A Different Approach for the Planning of Dental Caries Preventive Programs

Carlos Wagner Werner BDS, MSc, DrPH / Assistant Professor in the Community Dentistry Department, University of Texas / Health Sciences Center at San Antonio


Abstract The aim of this project was to develop a Knowledge-Based Expert System (KBES) prototype to guide the planning of dental caries preventive programs. While the final choice must include judgement on the part of local decision makers, such computer program may offer a valuable framework to guide the process of choice. The KBES prototype has been called Preventive Program Model (PPM) and was developed using the C Language Integrated Production System (Clips).

It has three modules:

1) Strategies recommended for fluoride-, diet, and sealant-based programs,

2) Efficacy/effectiveness of dental caries preventive measures, and

3) An estimation model for the effectiveness of the sealant-based program.

A pilot validation questionnaire distributed to 18 dental experts tested the program input, output and the reasoning process. Fifty percent of respondents (n=13) agreed that there was enough input information to recommend particular strategies, and 80% of the PPM outcomes for Module 1 were equivalent to the answers given by at least 50% of such respondents. In Module 3, the PPM estimation on the effectiveness of pit-and-fissure sealants gave comparable results to the expert estimation. Such results suggest the potential use of KBES in dental public health programs.

Introduction The scientific literature provides evidence on the efficacy and effectiveness of various caries prevention methods and regimens. Studies were carried out, however, in populations with diverse epidemiological profiles; thus, many important variables may not have been clearly stated or considered. These studies also may provide conflicting information. As a result, dental health officers frequently implement a particular preventive program without a clear idea of the likely impact of such program on the oral health of the particular community they seek to reach. If more precise estimates of potential results were available, such data, combined with data on costs, would be important for planning purposes. The planning process, however, would still demand a complex mix of skills and knowledge which often are not present, especially in developing countries ere the scientific or professional literature is limited. If information on the effectiveness of preventive dental measures were organized in such a way to guide the decision making process, local dental health officers would be more likely to make more effective decisions.

A review of the literature showed that no Knowledge-Based Expert Systems (KBES) has been reported with the task of guiding the planning of dental caries preventive programs. The development of a computer program based on KBES technology is an attempt to reduce the need for on-site consultants with wide-ranging expertise on the planning of dental public health programs.

The aim of this project is to develop a KBES prototype for the planning of community-based preventive dental programs. This prototype has been called the Preventive Program Model (PPM). Its main purpose is to guide oral health planers in choosing among alternative dental caries preventive programs based on locally derived data.

The development of a computer program such as this, demands the definition of the knowledge domain. Next sessions describe the knowledge representation used in the PPM. Following, the results of the pilot validation questionnaire will be shown as well the conclusions of the project.

Knowledge Representation The knowledge representation has been conceived in a three-module design:

1. Strategies recommended for fluoride-, diet-, and sealant-based programs;

2. Efficacy/effectiveness of dental caries preventive measures;

3. An estimation model for the effectiveness of the sealant-based program.

The PPM arbitrarily provides information for one specific age group (7-year-old children). The PPM is a working model that aims to demonstrate the feasibility of using a KBES for the planning of preventive dental programs. Use of the PPM in the field demands further expansion of the prototype.

Module 1: Strategies recommended for fluoride-, diet-, and sealant-based programs Module 1 guides the use to input the data relevant to the community where one wants to operate. If there is shortage of information, the user may make an "educated guess" and make arrangements to have the data in the future if possible. After local information is entered, a general preventive strategy is recommended. The strategies are based on the use fluoride-based products, diet-based regimens, and use of pit-and-fissure sealants. The user can limit the choice of strategies by choosing on the menu driven program, one or two of the three above mentioned choices. Recommendation may be for the use of the preventive program in the whole population or targeting a particular segment of the population (moderate and/or high risk). There is also the option of considering certain preventive program for one or more segments of the population. However, "recommended strategies" imply more need than "considered strategies".

The knowledge representation of Module 1 (Figure 1) is based on the combination of four parameters:

. Epidemiological profile of the population.

. Mean DMFT trend of the population.

. Level of fluoride in the water supplies.

. Level of refined carbohydrate consumption.

These parameters are grouped in order to determine the specific strategy for each community. The possible recommended and considered strategies are listed in the bottom part of Figure 1.

Module 2: Efficacy/effectiveness of dental caries preventive measures Module 2 provides information on the effectiveness of different preventive measures. It is an on-line bibliographical facility with a literature review of each preventive measure commercially available. It also presents a list of the references used, a graphic representation of results and the background history of the pioneering work and first clinical or field trials.

The knowledge representation for this module (Figure 2) allows for information retrieval and prompt access to a literature review on the effectiveness of preventive measures.

The dental caries preventive measures described in this module are: fluoridation of water supplies, fluoridation of school water supplies, salt fluoridation, topically applied fluoride solutions and gels, fluoride drops, and tablets as dietary supplements, fluoride mouthrinses, fluoride dentifrice, fluoride varnish, slow release fluoride devices, fluoride in beverages, pit-and-fissure sealants, and diet programs.

Module 3: An estimation model for the effectiveness of the sealant based program Module 3 develops a predictive model to estimate the likely impact of using pit-and-fissure sealants on first and second permanent molars. Following the results from the sealant strategy suggested in Module 1 of the PPM, the user is guided to estimate the possible range of effectiveness of sealants given by the percent of sound occlusal surfaces that would be decayed has sealants not been placed the in first permanent molars of 7-year-old children. The PPM asks the user to choose a time frame between one and ten years to estimate the sealant effectiveness. Based on the mean DMFT estimation for the population for the amount of years chosen, the PPM calculates the likely percent of effectiveness of the pit-and-fissure sealants. Figure 3 illustrates graphically the different stages of the sealant effectiveness estimation model, 5 years after the first permanent molars of 7-year-old children were sealed.

Validation Questionnaire A pilot validation questionnaire distributed to 18 dental experts that had a 70% response rate (13/18) examined the program input, output and different aspects of the reasoning process. The results of the questionnaire are listed next:

Program Input: Fifty percent of the experts who answered the questionnaire agreed that there is sufficient input data to plan community-based dental preventive programs. The prior history of dental caries in primary dentition and the population socioeconomic status however, were systematically suggested as desirable or essential information.

Program Output: Eight percent of the PPM outcomes for Module 1 was equivalent to the answers given by at least 50% of those who agreed that enough information had been provided to recommend a preventive strategy. The outcome for Module 3, on the sealant effectiveness varied according to the different scenarios provided in the questionnaire.

The expert response used a particular estimation model that when used by the PPM gave similar results for the lower caries prevalence populations and more accurate results for the higher caries prevalence populations. This finding suggests that the use of KBES may calibrate expert advises as well as predict results in a more accurate manner.

Reasoning Process: Some specific assumptions were used in the reasoning process of Modules 1 and 3. Two of such assumptions were tested in the pilot questionnaire: (1) necessary information for individual risk assessment and (2) the boundaries chosen for determining low-, moderate-and high-risk 7-year-old children. Special attention was given to the determination of the boundaries between low-and moderate - risk in lower age groups. Experts suggested the use of prior history of dental caries in primary dentition, socioeconomic status and microbiological tests.

Conclusion The PPM is the first KBES prototype known to have been developed for dental public health programs. It attempts to organize the necessary knowledge domain for the planning of preventive dental programs.

The main purpose in the development of expert systems for dental public health programs is to provide a simple and accessible system for use in communities with different epidemiological profiles and lack of access to experts. Expert systems also should provide state-of-the-art information to lead local decision makers into making more intelligent decisions. An expert system that covers strategies issue such as fluoride-, diet-, and sealant-based programs also helps to combine expertise sometimes not provided by just one person.

Finally, the search for reaching a consensus among experts may lead to a review on the role of dental public health researchers in providing scientifically based guidance to dental public health officers working in the field.

Acknowledgment Supported by the fellowship provided by the Brazilian National Research Council (CNPq).

I also would like to thank Dr. Steve Eklund, Dr Bryan Burt, Dr. Iris Tommelein and Dr. Paul Lang from the University of Michigan for their guidance and support.

Last but not least, I thank the Cedros Network for their continuing support and help.

Mailing address:

University of Texas / Health Science Center

San Antonio, Texas, USA

7703 Floyd Cure Drive

Fax: (001) (210) 567-3211


[CEDROS Home Page] [Previous Article] [Next Article] [List of Articles]

CEDROS


Copyright 1995 CEDROS Network. All Rights Reserved.