Olast Arrizibita Iriarte
Research Subjects
Research Interests
Olast Arrizibita Iriarte has a total of 34 co-authors. Here are the top co-authors by number of shared publications:
- Javier Rodríguez
Department of Medical Oncology, Clínica Universidad de Navarra, 31008 Pamplona, Spain
3 shared publications - Onintza Sayar Beristain
Department of Mathematics and Statistic, NNBi, 31110 Noain, Spain
3 shared publications - Oskitz Ruiz Sarrias
Department of Mathematics and Statistic, NNBi, 31110 Noain, Spain
3 shared publications - Ángel Vizcay Atienza
Department of Medical Oncology, Clínica Universidad de Navarra, 31008 Pamplona, Spain
3 shared publications - Teresa Zumárraga Lizundia
Department of Medical Oncology, Clínica Universidad de Navarra, 31008 Pamplona, Spain
2 shared publications - Adriana Martinez-Lage
Department of Radiation Oncology, Clínica Universidad de Navarra, 31008 Pamplona, Spain
1 shared publication
Latest Publications
- Open Access
- 21 May 2026
- Published by Public Library of Science (PLoS) in PLOS Digital Health
- Vol. 5 (5) , e0001441
Optimization of artificial intelligence models for prediction of new-onset cardiovascular disease in patients with arterial hypertension
Advanced preventive strategies are needed to decrease the burden of cardiovascular disease (CVD). We aimed to develop a predictive tool to identify individuals at higher CVD risk and facilitate proactive interventions to improve clinical outcomes. This single-center retrospective study enrolled consecutive hypertensive subjects free of CVD at baseline and followed them up for a mean of 8.3 years. The primary outcome was new-onset CVD (ischemic heart disease, stroke or hospitalization due to heart failure). The 155-variable dataset was enriched by creating trend variables using statistical measures, Principal Component Analysis (PCA) and Latent Class Analysis (LCA). Then, an artificial intelligence (AI) XGBoost prediction algorithm was trained on 70% of the dataset and validated on the remaining 30%. XGBoost-based risk stratification was compared with risk stratification according to SCORE2. The 3,588 consecutive patients enrolled had a mean age of 54.2 ± 14 years, 53% were women. The incidence rate of new-onset CVD was 1.93 (95% CI: 1.78-2.09) per 100 patient-years. The XGBoost model incorporated 30 variables and achieved 86% ROC AUC, 81% sensitivity, and 78% specificity for predicting CVD. The number of antihypertensive drugs had the strongest predictive power within the model. SCORE2 classified at baseline only 32% of participants with a CV event in the follow-up as high or very-high risk, whereas the XGBoost model correctly identified 81% of them. AI-based modeling outperformed SCORE2 in predicting new-onset CVD in patients with hypertension, identifying the number of antihypertensive drugs as a key predictor and supporting the role of AI risk stratification in clinical practice to implement precision medicine.
- 21 May 2026
- Published by Public Library of Science (PLoS) in PLOS Digital Health
- Vol. 5 (5) , e0001441
Open AccessOptimization of artificial intelligence models for prediction of new-onset cardiovascular disease in patients with arterial hypertension
Advanced preventive strategies are needed to decrease the burden of cardiovascular disease (CVD). We aimed to develop a predictive tool to identify individuals at higher CVD risk and facilitate proactive interventions to improve clinical outcomes. This single-center retrospective study enrolled consecutive hypertensive subjects free of CVD at baseline and followed them up for a mean of 8.3 years. The primary outcome was new-onset CVD (ischemic heart disease, stroke or hospitalization due to heart failure). The 155-variable dataset was enriched by creating trend variables using statistical measures, Principal Component Analysis (PCA) and Latent Class Analysis (LCA). Then, an artificial intelligence (AI) XGBoost prediction algorithm was trained on 70% of the dataset and validated on the remaining 30%. XGBoost-based risk stratification was compared with risk stratification according to SCORE2. The 3,588 consecutive patients enrolled had a mean age of 54.2 ± 14 years, 53% were women. The incidence rate of new-onset CVD was 1.93 (95% CI: 1.78-2.09) per 100 patient-years. The XGBoost model incorporated 30 variables and achieved 86% ROC AUC, 81% sensitivity, and 78% specificity for predicting CVD. The number of antihypertensive drugs had the strongest predictive power within the model. SCORE2 classified at baseline only 32% of participants with a CV event in the follow-up as high or very-high risk, whereas the XGBoost model correctly identified 81% of them. AI-based modeling outperformed SCORE2 in predicting new-onset CVD in patients with hypertension, identifying the number of antihypertensive drugs as a key predictor and supporting the role of AI risk stratification in clinical practice to implement precision medicine.
- Open Access
- 16 March 2026
- Published by Springer Nature in European Child & Adolescent Psychiatry
- Vol. 35 (6) , 1947-1956
Demographic and socioeconomic factors influencing ADHD diagnosis in children and adolescents: a population-based nested case–control study
Attention deficit hyperactivity disorder (ADHD) is among the most commonly diagnosed psychiatric disorders in childhood, with an estimated prevalence in children and adolescents of 5.9% in Navarre (2019) and 6.8% in Spain (2012). However, demographic and socioeconomic factors influencing diagnosis remain insufficiently explored. The aim was to examine associations between demographic factors (month of birth, migration status, number of siblings) or socioeconomic variables (household income, residence type, school type) and ADHD diagnosis in children and adolescents. Population-based case-control study nested in a cohort of individuals born between 1991 and 2011 and followed from 2007 to 2019 using linked health and educational records from Navarre, Spain. Incident ADHD cases were age and sex-matched to three controls (1:3). Data were analyzed with multivariate conditional logistic regression. Lower family income (< €18,000/year) was associated with a 23% increased risk of ADHD diagnosis (OR = 1.23, 95%CI: 1.15–1.31). Large sized rural areas showed less likely diagnostic rates compared to urban residence (OR = 0.88, 95%CI: 0.82–0.93). The month of birth significantly influenced diagnosis likelihood, with those born later in the year showing increased odds, peaking at 69% higher odds in October-December (OR = 1.69, 95%CI: 1.55–1.83). Having ≥ 3 siblings reduced diagnosis likelihood by 17% compared to only children (OR = 0.83, 95%CI: 0.76–0.91). Migration status and school type interaction analysis revealed that migrant children in public schools had significantly lower ADHD diagnoses compared to non-migrant peers; this difference diminished in charter schools. This study confirms, in a large population cohort, the significant impact of socio-demographic factors on the likelihood of ADHD diagnosis, such as socio-economic status, month of birth, migrant status, number of siblings, and place of residence, and provides new insights into the role of school type and the interaction between school type and migrant status, this interaction representing a relatively understudied factor. These associations highlight disparities and potential diagnostic biases. Strategies to address these inequalities are essential for consistent ADHD diagnostic practices and equitable healthcare interventions.
- 16 March 2026
- Published by Springer Nature in European Child & Adolescent Psychiatry
- Vol. 35 (6) , 1947-1956
Open AccessDemographic and socioeconomic factors influencing ADHD diagnosis in children and adolescents: a population-based nested case–control study
Attention deficit hyperactivity disorder (ADHD) is among the most commonly diagnosed psychiatric disorders in childhood, with an estimated prevalence in children and adolescents of 5.9% in Navarre (2019) and 6.8% in Spain (2012). However, demographic and socioeconomic factors influencing diagnosis remain insufficiently explored. The aim was to examine associations between demographic factors (month of birth, migration status, number of siblings) or socioeconomic variables (household income, residence type, school type) and ADHD diagnosis in children and adolescents. Population-based case-control study nested in a cohort of individuals born between 1991 and 2011 and followed from 2007 to 2019 using linked health and educational records from Navarre, Spain. Incident ADHD cases were age and sex-matched to three controls (1:3). Data were analyzed with multivariate conditional logistic regression. Lower family income (< €18,000/year) was associated with a 23% increased risk of ADHD diagnosis (OR = 1.23, 95%CI: 1.15–1.31). Large sized rural areas showed less likely diagnostic rates compared to urban residence (OR = 0.88, 95%CI: 0.82–0.93). The month of birth significantly influenced diagnosis likelihood, with those born later in the year showing increased odds, peaking at 69% higher odds in October-December (OR = 1.69, 95%CI: 1.55–1.83). Having ≥ 3 siblings reduced diagnosis likelihood by 17% compared to only children (OR = 0.83, 95%CI: 0.76–0.91). Migration status and school type interaction analysis revealed that migrant children in public schools had significantly lower ADHD diagnoses compared to non-migrant peers; this difference diminished in charter schools. This study confirms, in a large population cohort, the significant impact of socio-demographic factors on the likelihood of ADHD diagnosis, such as socio-economic status, month of birth, migrant status, number of siblings, and place of residence, and provides new insights into the role of school type and the interaction between school type and migrant status, this interaction representing a relatively understudied factor. These associations highlight disparities and potential diagnostic biases. Strategies to address these inequalities are essential for consistent ADHD diagnostic practices and equitable healthcare interventions.
- Open Access
Neoadjuvant Chemoradiotherapy in Locally Advanced Gastric Adenocarcinoma: Long-Term Results and Statistical Algorithm to Predict Individual Risk of Relapse
Background: The purpose of this study was to evaluate the long-term outcomes of patients with locally advanced gastric adenocarcinoma (LAGC) intended to receive induction chemotherapy, chemoradiation and surgery and to develop an algorithm to estimate the individual risk of relapse in a population-based setting. Methods: Patients with LAGC (cT3-4 and/or N+) were retrospectively evaluated. A pathological response was graded according to the Becker criteria. The nodal regression grade was assessed by a 4-point scale (A–D). A comprehensive analysis of 155 individual patient variables was performed, and logistic regression (LR) was utilized to develop a predictive model for relapse risk. Results: From 2010 to 2024, 48 patients were analyzed. After a median follow-up of 49 months (range, 12–212), the 5-year actuarial PFS and OS rates were 44% and 48%, respectively. Four variables were identified as the most relevant features for training the LR model. Scores for the model accuracy, sensitivity and specificity (mean +/− sd) were 0.79 +/− 0.12, 0.74 +/− 0.221 and 0.88 +/− 0.14, respectively. For a validation dataset, the figures were 0.78, 0.88 and 0.73, respectively. Conclusions: This neoadjuvant strategy seems to correlate with a favorable long-term outcome in a subset of intestinal-type LAGA patients who achieve ypN0 features.
Open AccessNeoadjuvant Chemoradiotherapy in Locally Advanced Gastric Adenocarcinoma: Long-Term Results and Statistical Algorithm to Predict Individual Risk of Relapse
Background: The purpose of this study was to evaluate the long-term outcomes of patients with locally advanced gastric adenocarcinoma (LAGC) intended to receive induction chemotherapy, chemoradiation and surgery and to develop an algorithm to estimate the individual risk of relapse in a population-based setting. Methods: Patients with LAGC (cT3-4 and/or N+) were retrospectively evaluated. A pathological response was graded according to the Becker criteria. The nodal regression grade was assessed by a 4-point scale (A–D). A comprehensive analysis of 155 individual patient variables was performed, and logistic regression (LR) was utilized to develop a predictive model for relapse risk. Results: From 2010 to 2024, 48 patients were analyzed. After a median follow-up of 49 months (range, 12–212), the 5-year actuarial PFS and OS rates were 44% and 48%, respectively. Four variables were identified as the most relevant features for training the LR model. Scores for the model accuracy, sensitivity and specificity (mean +/− sd) were 0.79 +/− 0.12, 0.74 +/− 0.221 and 0.88 +/− 0.14, respectively. For a validation dataset, the figures were 0.78, 0.88 and 0.73, respectively. Conclusions: This neoadjuvant strategy seems to correlate with a favorable long-term outcome in a subset of intestinal-type LAGA patients who achieve ypN0 features.
- Open Access
- 15 August 2024
- Published by MDPI AG in Biomedicines
- Vol. 12 (8) , 1859
Neoadjuvant Statistical Algorithm to Predict Individual Risk of Relapse in Patients with Resected Liver Metastases from Colorectal Cancer
(1) Background: Liver metastases (LM) are the leading cause of death in colorectal cancer (CRC) patients. Despite advancements, relapse rates remain high and current prognostic nomograms lack accuracy. Our objective is to develop an interpretable neoadjuvant algorithm based on mathematical models to accurately predict individual risk, ensuring mathematical transparency and auditability. (2) Methods: We retrospectively evaluated 86 CRC patients with LM treated with neoadjuvant systemic therapy followed by complete surgical resection. A comprehensive analysis of 155 individual patient variables was performed. Logistic regression (LR) was utilized to develop the predictive model for relapse risk through significance testing and ANOVA analysis. Due to data limitations, gradient boosting machine (GBM) and synthetic data were also used. (3) Results: The model was based on data from 74 patients (12 were excluded). After a median follow-up of 58 months, 5-year relapse-free survival (RFS) rate was 33% and 5-year overall survival (OS) rate was 60.7%. Fifteen key variables were used to train the GBM model, which showed promising accuracy (0.82), sensitivity (0.59), and specificity (0.96) in predicting relapse. Similar results were obtained when external validation was performed as well. (4) Conclusions: This model offers an alternative for predicting individual relapse risk, aiding in personalized adjuvant therapy and follow-up strategies.
- 15 August 2024
- Published by MDPI AG in Biomedicines
- Vol. 12 (8) , 1859
Open AccessNeoadjuvant Statistical Algorithm to Predict Individual Risk of Relapse in Patients with Resected Liver Metastases from Colorectal Cancer
(1) Background: Liver metastases (LM) are the leading cause of death in colorectal cancer (CRC) patients. Despite advancements, relapse rates remain high and current prognostic nomograms lack accuracy. Our objective is to develop an interpretable neoadjuvant algorithm based on mathematical models to accurately predict individual risk, ensuring mathematical transparency and auditability. (2) Methods: We retrospectively evaluated 86 CRC patients with LM treated with neoadjuvant systemic therapy followed by complete surgical resection. A comprehensive analysis of 155 individual patient variables was performed. Logistic regression (LR) was utilized to develop the predictive model for relapse risk through significance testing and ANOVA analysis. Due to data limitations, gradient boosting machine (GBM) and synthetic data were also used. (3) Results: The model was based on data from 74 patients (12 were excluded). After a median follow-up of 58 months, 5-year relapse-free survival (RFS) rate was 33% and 5-year overall survival (OS) rate was 60.7%. Fifteen key variables were used to train the GBM model, which showed promising accuracy (0.82), sensitivity (0.59), and specificity (0.96) in predicting relapse. Similar results were obtained when external validation was performed as well. (4) Conclusions: This model offers an alternative for predicting individual relapse risk, aiding in personalized adjuvant therapy and follow-up strategies.
- Open Access
Predicting Severe Haematological Toxicity in Gastrointestinal Cancer Patients Undergoing 5-FU-Based Chemotherapy: A Bayesian Network Approach
Purpose: Severe toxicity is reported in about 30% of gastrointestinal cancer patients receiving 5-Fluorouracil (5-FU)-based chemotherapy. To date, limited tools exist to identify at risk patients in this setting. The objective of this study was to address this need by designing a predictive model using a Bayesian network, a probabilistic graphical model offering robust, explainable predictions. Methods: We utilized a dataset of 267 gastrointestinal cancer patients, conducting preprocessing, and splitting it into TRAIN and TEST sets (80%:20% ratio). The RandomForest algorithm assessed variable importance based on MeanDecreaseGini coefficient. The bnlearn R library helped design a Bayesian network model using a 10-fold cross-validation on the TRAIN set and the aic-cg method for network structure optimization. The model’s performance was gauged based on accuracy, sensitivity, and specificity, using cross-validation on the TRAIN set and independent validation on the TEST set. Results: The model demonstrated satisfactory performance with an average accuracy of 0.85 (±0.05) and 0.80 on TRAIN and TEST datasets, respectively. The sensitivity and specificity were 0.82 (±0.14) and 0.87 (±0.07) for the TRAIN dataset, and 0.71 and 0.83 for the TEST dataset, respectively. A user-friendly tool was developed for clinical implementation. Conclusions: Despite several limitations, our Bayesian network model demonstrated a high level of accuracy in predicting the risk of developing severe haematological toxicity in gastrointestinal cancer patients receiving 5-FU-based chemotherapy. Future research should aim at model validation in larger cohorts of patients and different clinical settings.
Open AccessPredicting Severe Haematological Toxicity in Gastrointestinal Cancer Patients Undergoing 5-FU-Based Chemotherapy: A Bayesian Network Approach
Purpose: Severe toxicity is reported in about 30% of gastrointestinal cancer patients receiving 5-Fluorouracil (5-FU)-based chemotherapy. To date, limited tools exist to identify at risk patients in this setting. The objective of this study was to address this need by designing a predictive model using a Bayesian network, a probabilistic graphical model offering robust, explainable predictions. Methods: We utilized a dataset of 267 gastrointestinal cancer patients, conducting preprocessing, and splitting it into TRAIN and TEST sets (80%:20% ratio). The RandomForest algorithm assessed variable importance based on MeanDecreaseGini coefficient. The bnlearn R library helped design a Bayesian network model using a 10-fold cross-validation on the TRAIN set and the aic-cg method for network structure optimization. The model’s performance was gauged based on accuracy, sensitivity, and specificity, using cross-validation on the TRAIN set and independent validation on the TEST set. Results: The model demonstrated satisfactory performance with an average accuracy of 0.85 (±0.05) and 0.80 on TRAIN and TEST datasets, respectively. The sensitivity and specificity were 0.82 (±0.14) and 0.87 (±0.07) for the TRAIN dataset, and 0.71 and 0.83 for the TEST dataset, respectively. A user-friendly tool was developed for clinical implementation. Conclusions: Despite several limitations, our Bayesian network model demonstrated a high level of accuracy in predicting the risk of developing severe haematological toxicity in gastrointestinal cancer patients receiving 5-FU-based chemotherapy. Future research should aim at model validation in larger cohorts of patients and different clinical settings.