Predicting Periodontal Treatment Success Using Machine Learning in Periodontitis Patients

Development of a Machine Learning-Assisted Model for Predicting Post-Periodontal Treatment Success and Individual Risk Analysis: A Retrospective Cohort Study

The aim of this study is to develop a clinical decision-support model capable of predicting the optimal periodontal treatment option at the individual patient level by utilizing a multidimensional dataset composed of clinical periodontal parameters, radiographic findings, implemented treatment modalities, and demographic characteristics. In this context, the study seeks to strengthen personalized treatment planning by identifying the most effective therapeutic approach for individuals presenting for periodontal care.

Study Overview

• Status: Completed
• Conditions: Periodontitis
• Intervention / Treatment: Procedure: Phase-1 Periodontal Therapy; Procedure: Conventional Flap Surgery; Procedure: Regenerative Flap Surgery

Detailed Description

Periodontitis is a highly prevalent, complex, and multifactorial chronic inflammatory condition affecting the gingiva, periodontal ligament, cementum, and alveolar bone, in which a microbially driven, host-mediated immune-inflammatory response ultimately results in periodontal attachment loss and alveolar bone resorption.

The diagnosis of periodontitis relies on a thorough clinical and radiographic assessment of the periodontal tissues. The characterization of the disease commonly includes the number and proportion of teeth presenting probing pocket depths exceeding specific thresholds (most frequently >4 mm with bleeding on probing and ≥6 mm), the number of teeth lost due to periodontitis, the number of teeth exhibiting intrabony defects, and the number of teeth with furcation involvement, all of which serve as clinically meaningful indicators.The classification update released in 2017 transformed periodontal diagnostics by adopting a stage-and-grade system, which allows clinicians to evaluate disease severity, anticipated progression, and the likelihood of future relapse with greater precision. Although these criteria effectively identify established disease, they primarily reflect historical tissue destruction and provide limited insight into current disease activity or future progression. Consequently, there is growing interest in more sensitive, specific, and non-invasive diagnostic approaches that can improve early detection and prognostic accuracy. However, despite this structured approach, clinicians still face difficulties because periodontitis develops through a highly variable interplay of host immune function, microbial imbalance, genetic factors, and lifestyle or environmental influences. Periodontal treatment is generally divided into non-surgical and surgical approaches. Non-surgical therapy (Phase I treatment) primarily includes supragingival and subgingival debridement procedures, focusing on the removal of dental calculus and the smoothing of root surfaces. In some cases, however, due to disease progression, anatomical complexities, or patient-specific host factors, surgical intervention (Phase II treatment) may become necessary. Surgical treatment options include flap surgery, resective procedures, and regenerative techniques. Although clinical parameters such as probing pocket depth, bleeding on probing, and clinical attachment level can guide the decision to transition from Phase I to Phase II therapy, this decision is often individualized and based on the clinician's expertise and patient-specific considerations.

Artificial intelligence represents a field within computer science dedicated to creating systems that can perform tasks typically requiring human cognitive abilities-often more rapidly and with greater precision. Within this field, machine learning (ML) involves developing statistical algorithms that can analyze and categorize data or images, as well as predict risks and outcomes using a variety of computational techniques.Artificial intelligence (AI) applications in periodontology are extensive and primarily focused on enhancing disease classification, diagnosis, and treatment planning. In treatment planning, AI facilitates the segmentation of periodontal structures, allowing clinicians to visualize and simulate surgical outcomes in a virtual environment.

The integration of an AI-based model may maximize the likelihood of achieving successful periodontal outcomes and guide periodontists in selecting the most appropriate treatment plan. In light of this potential, the aim of the present study is to develop a decision-support model capable of predicting the optimal periodontal treatment option at the individual patient level by using advanced machine learning algorithms on a multidimensional dataset comprising clinical periodontal parameters, radiographic data, applied treatment modalities, and demographic information. By doing so, the study seeks to support personalized treatment planning by identifying the most effective therapeutic approach for individuals undergoing periodontal therapy.

• Study Type: Observational
• Enrollment (Actual): 86

Contacts and Locations

Study Locations

• Turkey (Türkiye)
  • konyaaltı
    • Antalya, konyaaltı, Turkey (Türkiye), 07070
      • Akdeniz University

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Child; Adult; Older Adult
• Accepts Healthy Volunteers: No

• Sampling Method: Non-Probability Sample

Study Population

The study population consists of patients diagnosed with periodontitis who applied to the Akdeniz University Faculty of Dentistry, Department of Periodontology clinic between 2021 and 2025. Participants were selected from patients who underwent Phase-1 periodontal therapy or surgical periodontal procedures (conventional or regenerative flap surgery) and had complete clinical and radiographic records in the Hospital Information Management System database.

Description

Inclusion Criteria:

1. A confirmed diagnosis of periodontitis supported by complete clinical and radiographic records.
2. Completion of both first and second stage periodontal therapy, including oral hygiene instruction and scaling root planning (SRP).
3. Attendance at a minimum of one follow-up visit after completion of initial and secondary periodontal treatment.
4. Persistence of bleeding on probing (BoP), probing pocket depth (PPD) ≥5 mm, or worsening periodontal parameters despite adequate oral hygiene, resulting in an indication for periodontal surgery.
5. Availability of detailed documentation for each tooth, including the type of active treatment performed and corresponding post treatment records.
6. Patients with a previous history of cancer were eligible provided that chemotherapy or radiotherapy had been completed and medical clearance for periodontal treatment had been obtained

Exclusion Criteria:

1. Demographic, clinical, or radiographic data were incomplete.
2. Systemic conditions contraindicating periodontal treatment were present.
3. Pregnancy or breastfeeding at the time of periodontal treatment.
4. Ongoing chemotherapy or radiotherapy.
5. Current use of bisphosphonate therapy.
6. Presence of an immunocompromised condition.
7. Acute systemic illness or active infection at the time of evaluation.

Study Plan

How is the study designed?

Number of groups / cohorts

3

Cohorts and Interventions

Group / Cohort	Intervention / Treatment
Phase-1 Periodontal Therapy; Patients who received non-surgical periodontal treatment consisting of oral hygiene instructions, scaling, and root planing (SRP).	Procedure: Phase-1 Periodontal Therapy; Non-surgical periodontal treatment consisting of scaling and root planing (SRP) under local anesthesia, along with oral hygiene instructions
Conventional Flap Surgery; Patients who underwent traditional periodontal flap surgery (access flap) following unsuccessful non-surgical therapy to reduce pocket depth.	Procedure: Conventional Flap Surgery; Periodontal access flap surgery performed for subgingival debridement and pocket depth reduction in cases unresponsive to Phase-1 therapy.
Regenerative Flap Surgery; Patients who underwent periodontal surgery involving regenerative materials (bone grafts, membranes, or enamel matrix derivatives) for the treatment of intrabony defects.	Procedure: Regenerative Flap Surgery; Surgical intervention utilizing regenerative materials such as bone grafts or barrier membranes for the treatment of periodontal intrabony defects.

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Clinical Success of Periodontal Treatment	Binary classification (Success/Failure) of treatment outcomes based on clinical parameters. Success is defined as a reduction of at least 2mm in probing pocket depth (PPD) and a gain in clinical attachment level (CAL) at the 6-month follow-up compared to baseline.	Baseline and 6 months post-treatment.

Collaborators and Investigators

• Sponsor: Akdeniz University

Study record dates

Study Major Dates

• Study Start (Actual): August 2, 2025
• Primary Completion (Actual): January 31, 2026
• Study Completion (Actual): January 31, 2026

Study Registration Dates

• First Submitted: March 17, 2026
• First Submitted That Met QC Criteria: March 17, 2026
• First Posted (Actual): March 20, 2026

Study Record Updates

• Last Update Posted (Actual): March 25, 2026
• Last Update Submitted That Met QC Criteria: March 23, 2026
• Last Verified: March 1, 2026

More Information

Terms related to this study

• Keywords: machine learning; periodontitis; clinical decision support system
• Additional Relevant MeSH Terms: Periodontal Diseases; Mouth Diseases; Stomatognathic Diseases; Periodontitis
• Other Study ID Numbers: TBAEK-608; TDH-2025-6983 (Other Identifier: BAP)

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No