Resumen de Precision dental medicine for implant-supported oral rehabilitation: proposal of a prediction tool for the success of implants
The growth of population and of increased lifespan has meant that more people are looking for treatments and solutions for lost teeth, resulting in an increased demand for bone regeneration treatments and oral rehabilitation techniques for elderly patients with specific health conditions. Patient-related conditions, such as smoking habits, poor oral hygiene, infectious processes, systemic diseases (osteoporosis, diabetes mellitus), and drugs that affect bone metabolism, might influence the progress of bone regeneration and, consequently, the osseointegration of dental implants. In addition, factors related to the surgical and prosthetic phase, as well as the inherent characteristics of dental implants. Therefore, information about the rehabilitation, including the implant system used, fixation method, and abutment used, is needed. Patient history and radiographic examination provide information that allows the clinician to identify the implant system. The development of methodologies able to integrate all the factors and predictors is possible with the use of artificial intelligence (AI). These strategies support the prognosis of the implant, predicting eventual clinical conditions such as early bone loss, mucositis, or periimplantitis.
The scientific evidence, as well as the assessment tools used in contemporary practice, has been based on clinical, analytical, and radiographic parameters which provide the clinician with limited therapeutic guidelines to deal with the multifactorial complexity of the implantsupported rehabilitation procedures. Furthermore, for diagnosing and staging peri-implant disease, such methods can only register the actual tissue destruction rather Proposal of a prediction tool for the success of implants than current disease activity. Moreover, those conventional strategies do not consider systemic conditions, which may influence the local immunological response, either around a tooth area (periodontitis) or around a dental implant area (peri-implantitis). Currently, the role of pathogens and their influence on periodontal and peri-implant diseases have been well described and it has been reported that oral dysbiotic status is necessary to trigger these pathologies. This understanding has allowed the identification and confirmation of several individual conditions such as risk factors with immunological impact.
The Omics methodology (i.e. the term "omics" derives from the Greek word "omnis," meaning "all" or "complete," and is used to describe the holistic and systematic study of various biological components) is key to the introduction of precision medicine into dentistry, especially in the field of implant-supported oral rehabilitation, because it can adapt the procedure to follow considering the patients biological, social, and lifestyle characteristics. A major goal is to reduce diagnostic mistakes, to develop results, to avoid unnecessary collateral effects, and to clarify why one individual can develop peri-implantitis and others with similar conditions did not.
Objectives This doctoral thesis aims to take the first steps towards the creation of a protocol to be followed in cases of implant-supported oral rehabilitation, which complies with the assumptions of precision medicine. Considering the aim and transversality of precision medicine, it is imperative to create protocols that aim to respond to its assumptions, ideally through the application of AI algorithms and omics methodologies such as biomarkers. The specific objectives of this thesis were: 1) to review the literature on bioinformatics (artificial intelligence and omic sciences), addressing the state of the art of how its have been used to predict the success of dental implants; 2) to review how the molecular point-of-care (PoC) tests currently available can help in the early detection of peri-implant diseases; 3) to identify a test kit commercially available and approved on European Union that are already been validated to function with peri-implantitis biomarkers and that use oral fluid to diagnose; 4) to investigate dentists' perception of the implementation of a tool to support peri-implant risk assessment; 5) to create a usability test to identify improvements that can be made to the IDRA tool and, 6) to create a proposal tool to predict the success of dental implants.
Methods A bibliographic review was made in PubMed and Web of Science respecting the methodology described in the Preferred Reporting Items for Systematic reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) checklist. The focal question was how are bioinformatics being used in the field of oral implantology as a predictive tool to ensure implant success A second search strategy was created to answer the question: How the molecular point-of-care tests currently available can help in the early detection of peri-implant diseases and throws light on improvements in point of care diagnostics devices The methodology included applying a search strategy, defining inclusion and exclusion criteria, and retrieving studies; selecting studies; extract relevant data; and performing tables to summarize the results. Searches of PubMed and Web of Science were performed to gather literature published until September 2022.
A qualitative study was performed to explore dentists perceptions toward the implementation of a dental informatics risk assessment tool. The Implant Disease Risk Assessment Tool (IDRA) was presented to a convenience sample of seven dentists Proposal of a prediction tool for the success of implants working in a university clinic, who were asked to use IDRA with the information of three clinical cases whilst thinking aloud and then fill the System Usability Scale (SUS). A semi-structured interview technique was used with audio record to allow free expression of participants perceptions related to the IDRA. The interviews information was categorized and analyzed by the authors.
Results In the first review, three articles discussed bioinformatic models that integrated AI algorithms into established identification and quantification protocols, which are often used in Omics sciences. A total of 13 articles underlined the development of different AI algorithms, for example, machine learning, deep learning, and convolutional neural network to support clinical decision and raising precision and accuracy levels of the rehabilitation process. Of these, 6 studies developed AI models for implant type recognition. Most of the articles identified used AI algorithms as a clinical support tool, as opposed to the articles which applied bioinformatic strategies by combining knowledge from AI algorithms with Omics expertise. The conventional criteria currently used as a technique for the diagnosis and monitoring of dental implants are insufficient and have low accuracy. Models that apply AI algorithms combined with precision methodologies and biomarkers are extremely useful in the creation of precision medicine, allowing medical dentists to forecast the success of the implant. Tools that integrate the different types of data, including imaging, molecular, risk factor, and implant characteristics, are needed to make a more accurate and personalized prediction of implant success.
At the second review it was found, the PerioSafe® PRO DRS (dentognostics GmbH, Jena) and ImplantSafe® DR (dentognostics GmbH, Jena) ORALyzer® test kits, already used clinically, can be a helpful adjunct tool in enhancing the diagnosis and prognosis of periodontal/peri-implantar diseases. With the advances of sensor technology, the biosensors can perform daily monitoring of dental implants or periodontal diseases, making contributions to personal healthcare and improve the current status quo of health management and human health. More and more emphasis is given to the role of biomarkers in diagnosing and monitoring periodontal and peri-implant diseases. By combining these strategies with traditional protocols, professionals could increase the accuracy of early detection of peri-implant and periodontal diseases, predicting disease progression, and monitoring of treatment outcomes.
Regarding the usability test of the IDRA tool, to our knowledge, this was the first study conducted to develop a qualitative usability test of IDRA, evaluating the effectiveness, efficiency, and users¿ satisfaction. There were more variations in responses the greater the degree of complexity of the clinical case. Generally, the participants classified the tool as good, getting usability values of 77,2 (SD 19,8) and learnability 73,2 (SD 24,5). Four additional factors should be considered to improve IDRA tool: 1) considering the relation between contour angle and peri¿implant tissue height; 2) automatic periodontal classification in the IDRA tool after completing the periodontogram in the clinical software; 3) presentation of a flow chart to assist therapeutic decisions alongside the final score defined by the IDRA tool; 4) integrating of precision tests such as Implantsafe® DR (dentognostics gmbh, Jena) and Oralyzer®(dentognostics gmbh, Jena).
Etiology and pathogenesis of peri-implant diseases is multifactorial. These tools must follow a natural integration to be easily applied in a clinical setting. It is important to study their usability from the clinicians point of view, evaluating the effectiveness, efficiency, and users satisfaction.
Proposal of a prediction tool for the success of implants Conclusions Based on these findings, it is possible to create a proposal tool that will integrate the assumptions of precision medicine, incorporating updated strategies to support the diagnosis and predict the dental implant success. This proposal tool can be seen as an eventual update to IDRA, since it was mentioned by the authors that if additional factors become evident from the literature, modifications of the diagram may be appropriate.
The proposed tool created is called the Implant Failure Prediction Tool IFPT. The IFPT is not yet translated into digital format, it only exists as a concept design.
Currently, this tool has all the conditions to be used to assist in the early diagnosis of peri-implant diseases, namely mucositis and peri-implantitis, through ImplantSafe® test kits. Its completion and risk calculation follows the same rationale of IDRA. However, to improve the doctor-patient communication and to make it easier for the patient to understand and follow up his/her own case, the result provided by the IFPT is given as traffic signal, besides the written indication of the risk of developing a peri-implant disease. Thus, from the patient's point of view, the greenish the diagram is, the more possibility of implant success patient has. If a yellow vector appears, it means that the patient should modulate his or her behavior to change it to the green level; the more reds appear in the diagram, the higher the risk of developing peri-implant disease.
In the foreseeable future, it will function as an individualized tool that will accurately predict the success of the dental implant. Currently, what is within our reach is to start creating a diagnostic and prognostic model. Defining a longitudinal study methodology that allows the loading of as many clinical cases as possible into the IFPT ("inputs") and, through the follow-up of these patients, to identify/diagnose possible clinical outcomes of peri-implantitis, mucositis or peri-implant health ("outcomes") over time. In this way, as the data are processed by means of IA algorithms, the variables/predictors with more significance for the determination of the implant failure may be identified and, at the same time, their respective weights in the predicting algorithm. This methodology will be detailed throughout the next topic of the discussion. This tool will use AI algorithms, namely artificial neural networks technology, allowing the tool to accumulate different functions as it is used. Artificial neural networks are highly flexible models and have been used in medicine to explore relationships between various physiological variables and to build predictive models. In this way it is possible to define an algorithm capable of indicating with accuracy and precision treatment response.
