In everyday life, gum diseases rarely make headlines. They lack the drama of oncology or the spectacle of breakthroughs in cardiology. And yet, as chronic conditions, they rank among the most widespread inflammatory diseases in the world. According to U.S. data, nearly half of adults over thirty have periodontitis, and almost 8% present with its advanced forms. A disease that begins with innocent bleeding during brushing may, over time, lead to tooth loss, pain, difficulty eating, and an increased risk of heart disease and diabetes.

Modern periodontology faces a pressing question: how do we predict the future of an individual patient? How, in an ocean of personal factors—genetics, lifestyle, anatomy, microbiome—do we determine who needs urgent intervention and who requires only more attentive prevention?

This question has long driven the development of periodontal risk assessment tools. The best known, PRA, is based on six criteria and offers a three-tier risk structure. For some, this is sufficient; for others, too general. The global classification of periodontal diseases was updated only in 2018, after nearly two decades without systemic change. Meanwhile, dentistry and personalized medicine have leaped forward. The field was waiting for a tool more adequate to this new reality.

Enter GF-PeDRA©, an algorithm developed in St. Louis by Gustavo Vicentis de Oliveira Fernandes and Juliana Campos Hasse Fernandes — a system that attempts to gather all key risk factors into a single dynamic picture, something akin to an “oral climate map.”

Eighteen Vectors, One Patient

When a clinician looks at GF-PeDRA© results, they don’t see a sterile table. They see an octagon with eighteen arms, each representing a different dimension of periodontal health: pocket depth, bone loss, bleeding on probing, furcations, systemic diseases, smoking, presence of implants and their complications, and even whether bone loss is vertical or horizontal. It is like viewing a patient not through a magnifying glass but through a kaleidoscope—each rotation shifting the interpretation of risk.

For many dentists, this visualization is intuitive: instead of navigating a spreadsheet, they face an image. The larger the filled polygon, the worse the prognosis. The algorithm converts this area into a percentage score. The risk thresholds work almost like a weather forecast:

0–9% — good prognosis
10–24% — fair
25–37% — poor
38–49% — questionable
≥50% — hopeless

It is the first tool to translate patient complexity into a precise, comprehensible language of numbers and shapes.

In conversation with us, Gustavo Fernandes, Associate Professor and Assistant Director in Periodontics, Still University - School of Dentistry & Oral Health notes that the greatest change after implementing GF-PeDRA© occurred not in charts, but on the other side of the dental chair:

“The biggest change has been in improving patients’ understanding of their clinical condition,” he says. “The integrated, visual output makes it much easier to explain why someone falls into a certain risk category, and how systemic factors like diabetes or smoking interact with clinical findings. As a result, conversations are more structured, patients ask more focused questions, and they much more readily accept comprehensive treatment and maintenance plans. Treatment planning has become more consistent and transparent, but the real ‘game changer’ has been the quality and depth of communication with patients.”

221 Patients and a Perfect Match

In their study, the creators of GF-PeDRA© analyzed 221 university clinic patients. Data were first assessed by two independent periodontists. Their diagnoses agreed to a large extent (κ = 0.83), though not perfectly — 37 cases required additional discussion. In periodontology, this is normal: distinctions between disease stages are subtle, and radiographic interpretation or assessing long-term progression often relies on experience.

But when the same data were entered into GF-PeDRA©, something extremely rare in diagnostic research occurred: the algorithm’s agreement with the specialists’ final diagnosis reached κ = 1.0 — 100%.

This is the clinical equivalent of an orchestra playing without a single wrong note.

The mean GF-PeDRA© score in the sample was 28.64%, corresponding to a “poor” prognosis. The distribution was:

21.7% — good
19.5% — fair
19.5% — poor
30.8% — questionable
8.6% — hopeless

The last category does not mean a patient cannot be helped. It signals that the disease is so advanced that current treatment must be intensive and outcomes uncertain.

Why Do We Need a Better Risk Assessment System?

Periodontology, like much of medicine, is shifting from a reactive to a predictive model. Dentists no longer want to merely treat — they want to anticipate. They want to identify those who may lose periodontal bone within five years, even if they show no symptoms today.

Traditional risk tools, although reliable, have limitations: they consider few parameters, do not differentiate causes of tooth loss, don’t contextualize periodontal disease within systemic disease, ignore bone-loss patterns, and do not incorporate complex features such as occlusal trauma or chewing dysfunction.

GF-PeDRA© uses three times more parameters than classic PRA and integrates them with the global 2018 periodontal disease classification. This matters — the earlier classification had been in place for 19 years, during which radiology, microbiology, and digital diagnostics evolved dramatically.

The new tool enables:

more accurate distinction between chronic and aggressive forms,
visual identification of risk factors,
communication with patients who see their condition instead of hearing abstract terms,
longitudinal monitoring of progression.

It is not yet full artificial intelligence — but it is a step toward dentistry that not only analyzes the patient’s past but predicts their future.

Podcast: A Digital Tool in Periodontal Diagnosis

Can the Algorithm Predict the Future of the Periodontium?

Predictive medicine rests on one rule: an algorithm is only as good as the data it is fed. The developers acknowledge that GF-PeDRA© needs broader validation — larger, more representative populations, especially those with rare disease patterns such as necrotizing lesions or complex systemic conditions.

The study population was clinical and thus naturally more “diseased” than the general population — showing the tool’s potential but the need for broader testing.

Still, the fact that the algorithm perfectly replicated two experienced clinicians’ diagnoses is no coincidence. It means its logic, weights, and parameter structure align with real-world clinical decision-making.

Prof. Fernandes views this as a beginning:

“The next logical step is a prospective cohort study with at least two to five years of follow-up. We need to track changes in clinical attachment level (CAL), radiographic bone loss (RBL), tooth loss due to periodontitis (TLP), and disease recurrence. We plan to include more than 2,000 participants, stratified by baseline risk category, to allow robust subgroup analyses and model calibration. Endpoints are very specific: CAL, RBL and bone loss pattern, pocket recurrence, bleeding on probing, TLP, stabilization or progression of stage/grade, need for reintervention, and associations with systemic diseases and other risk factors. Such a project will show whether GF-PeDRA© truly predicts disease course — and whether its low/medium/high thresholds need adjustment.”

Can GF-PeDRA© predict whether a tooth will remain for five more years? We do not know yet. But we know it signals risk with high precision — and that is a value in itself.

An Algorithm With Limitations

Even the best model cannot replace clinical judgment. The creators emphasize that some parameters require experience: radiograph interpretation, furcation assessment, distinguishing bone loss due to periodontitis from that caused by prosthetic crowns. The algorithm also does not yet include genetic or microbiological data — though the system is ready for such extensions.

Still, in its current form, GF-PeDRA© accomplishes something previously missing: it translates disease complexity into a clear, repeatable decision framework.

Will GF-PeDRA© revolutionize periodontology? It is too early to say. But one thing is certain: in a world increasingly driven by data, even everyday diseases like periodontitis deserve technology that can grasp their full complexity.

And if the algorithm helps us see more than red gums and radiographic bone levels, it is a step toward medicine that not only treats, but predicts.

In the era of digital dentistry, tools like this may become standard — like dermatoscopes in dermatology or cardiovascular risk scores in cardiology.

D. Sikora

New algorithm/tool used to achieve the periodontal risk assessment, diagnosis and prognosis (GF-PeDRA©):A clinical study with 221 patients

FAQ: GF-PeDRA©: Diagnosis, Risk Assessment and Prognosis in Periodontology

What is the primary purpose of the GF-PeDRA© algorithm?

GF-PeDRA© was developed to provide a comprehensive, automated periodontal and peri-implant diagnosis, integrated risk assessment, and treatment prognosis. It compares automated outputs with specialist clinical diagnoses and introduces a new scoring system that establishes cut-off values for prognostic categories based on an 18-parameter assessment model.

What parameters does GF-PeDRA© evaluate to generate periodontal risk and prognosis?

The tool incorporates 18 systemic and clinical variables, including probing depth, clinical attachment loss, radiographic bone loss, bleeding on probing, tooth loss, disease progression, biofilm accumulation, smoking, diabetes, furcation involvement, peri-implant disease, bone loss pattern, and additional systemic conditions. These factors are combined into an octadecagon visual model, producing a percentage-based score that determines prognosis.

How accurate was GF-PeDRA© compared to professional periodontal diagnoses?

GF-PeDRA© demonstrated perfect agreement (κ = 1.0) with the final professional diagnoses in the study sample of 221 patients. After independent evaluation by two clinicians and discussion of discrepant cases, the algorithm matched every established clinical diagnosis, confirming its high diagnostic reliability.

What were the main clinical outcomes observed in the study population?

Among the 221 patients, 28 were periodontally healthy, 55 had plaque-induced gingivitis, and 138 had periodontitis across all stages and grades. The mean GF-PeDRA© score was 28.64%, and prognosis categories were distributed as follows:

Good: 21.73%
Fair: 19.46%
Poor: 19.46%
Questionable: 30.77%
Hopeless: 8.60%

How does the new scoring system define prognosis categories?

The scoring system calculates the percentage area of the octadecagon based on parameter severity. Prognostic categories are defined as:

0–9%: Good
10–24%: Fair
25–37%: Poor
38–49%: Questionable
≥50%: Hopeless

This approach creates an objective framework for evaluating treatment predictability.

What limitations did the authors identify, and what future research is needed?

The study population included a high proportion of periodontitis cases, which may not fully represent the general population. The authors emphasize the need for larger cohorts and longitudinal studies to validate prognostic predictions, refine scoring weights, and assess long-term patient outcomes. They also note that collecting complete datasets for all 18 parameters may require more clinical time and experience.

Powerd by NotebookLM

Editor in Chief's Top Pick 2025

About the journal:

(Dent Med Probl) is an international, peer-reviewed, open access journal covering aspects of oral sciences and general medicine, published bimonthly by Wroclaw Medical University.

The journal is a leading dental scholarly journal in Eastern Europe founded in 1960 from the initiative of Prof. Tadeusz Owiński. It was originally called "Wrocławski Biuletyn Stomatologiczny". In 1965, the journal was renamed to "Wrocławska Stomatologia", and then the name was changed to "Dental and Medical Problems" in 2002.

Dental and Medical Problems is the first dentistry-profile scholarly journal in Poland and Eastern Europe in general which received a Journal Impact Factor (JIF) in the 2023 release of the Journal Citation Reports™. The current JIF is 3.9.

This material is based on the article:

New algorithm/tool used to achieve the periodontal risk assessment, diagnosis and prognosis (GF-PeDRA©): A clinical study with 221 patients

Gustavo Vicentis de Oliveira Fernandes, Juliana Campos Hasse Fernandes

Dental and Medical Problems

doi:10.17219/dmp/199583

Web. A. Maj

Photo: freepik.com, graphics from original article