PD29 Protocol For Rapid Reviews And Clinical Recommendations Using Artificial Intelligence Tools And Agile Processes - PD31 Counting Oncology Patients In The Brazilian Public Health System: A Comprehensive New Methodology

Sep 29, 2026

Uploaded by Miles

Rapid reviews are designed to preserve the methodological rigor of systematic reviews while reducing turnaround time, increasingly supported by artificial intelligence tools. An institutional protocol was developed from established guideline checklists and key systematic-review institutions, then iteratively refined through internal and external stakeholder evaluation. Agile sprint practices and five AI tools were tested, along with an online storage standard and a task checklist, enabling completion in about three months. A second study presents a reproducible, open-sourced SUS methodology using DATASUS claims, validated in pancreatic cancer patients aged 18–99 from 2008–2022.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages1

Reading Time3 min

AccessFree to download

PD29 Protocol For Rapid Reviews And Clinical Recommendations Using Artificial Intelligence Tools And Agile Processes

Introduction

Methods

Results

Conclusions

PD31 Counting Oncology Patients In The Brazilian Public Health System: A Comprehensive New Methodology

Introduction

Methods

Results

Conclusions

PD29 Protocol For Rapid Reviews And Clinical Recommendations Using Artificial Intelligence Tools And Agile Processes Leonardo Luna ([leoluna@hotmail.com](leoluna@hotmail.com)) Introduction: Rapid reviews are a subset of systematic reviews developed in the past decade to maintain methodological rigor while accelerating the review process. Increasingly, artificial intelligence tools are being utilized to enhance various aspects of these reviews, ensuring greater efficiency. Typically, rapid reviews support official recommendations by medical societies or healthcare institutions. Methods: Based on the Guidelines International Network-McMaster Guideline Development Checklist and publications from prominent institutions in systematic reviews, particularly the Cochrane Collaboration, the World Health Organization, McMaster University, and the Joanna Briggs Institute, an institutional protocol for rapid reviews was developed to support the creation of institutional clinical recommendations. This protocol was subjected to internal discussions within the Health Technology Assessment Unit of the National Institute of Cardiology, external evaluation by the Guidelines Committee of the Brazilian Society of Cardiology, and was refined throughout the development ofguidelines for these institutions. Agile processes and artificial intelligence tools were tested to enhance efficiency. Results: The development of one position statement, four clinical recommendations, and an institutional protocol facilitated refinement of the entire process. Weekly sprint methodology was incorporated, five artificial intelligence tools were integrated to support systematic review methods, an online storage standard was established, and a task checklist was created. These modifications enabled the completion of a rapid review in approximately three months. Conclusions: The globalization of health care and rapid system changes require medical institutions to quickly develop evidencebased recommendations. Rapid reviews accelerate access to quality knowledge. This paper highlighted tools tested and incorporated to streamline the process, ensuring efficiency while maintaining the quality and scientific rigor of the final document. PD31 Counting Oncology Patients In The Brazilian Public Health System: A Comprehensive New Methodology Patricia Klarmann Ziegelmann (patricia. [ziegelmann@ufrgs.br](ziegelmann@ufrgs.br)), Cintia Sayuri Kurokawa La Scala and Nelson Francisco Correa-Netto Introduction: In Brazil, the databases ofthe informatics department of the Brazilian Unified Health System (DATASUS) are key sources of oncology real-world data (RWD) . Despite challenges like data quality, transparent methods ensure robust real-world evidence for cancer care. This study proposed a reproducible and open-sourced methodology to count cancer cases in the Unified Health System (SUS) . Methods: The methodology counts cancer cases treated in the SUS using claims data available from DATASUS. The unique patient counting system employs patient ID, with deduplication for patients with more than one ID, using a key based on sex, ZIP code, International Classification of Diseases, 10th Revision (ICD-10) codes, diagnosis date, and calculated birth year. This new methodology, which includes data download, was implemented in an R software package. Data from DATASUS outpatient records (chemotherapy, immunotherapy, or radiotherapy) for patients aged 18 to 99 years with pancreatic cancer (ICD-10 code C25) diagnosed between 2008 and 2022 were used to validate the method. Results: A total of 254,240 outpatient claim records were analyzed: counting 31,425 unique patients by the proposed methodology and 1.3 percent of records where the patient ID represented duplication. Patients had a mean age of 60.9 years, 50.2 percent were female, and 82.8 percent were aged 50 years or older. Regarding the cancer stage, 3.5 percent had in situ, whereas 3.4 percent, 9.3 percent, 17.6 percent,

FAQs

What is the purpose of rapid reviews in this document?

Rapid reviews aim to maintain systematic-review methodological rigor while accelerating the review process, supporting official recommendations by medical societies or healthcare institutions.

How was the PD29 protocol developed and refined?

The PD29 protocol was built using established guideline checklists and systematic review publications, then refined through internal discussions and external evaluation from relevant guideline committees, while testing agile processes and AI tools for efficiency.

How does the PD31 methodology count oncology patients in Brazil’s public health system?

PD31 counts cancer cases treated in the SUS using DATASUS claims data and a unique patient counting approach with deduplication based on a key derived from sex, ZIP code, ICD-10 codes, diagnosis date, and calculated birth year.

Sep 29, 2026

Uploaded by Miles