Anxiety, Artifical Intelligence, Cancer, Depression
Conditions
Keywords
Artificial Intelligence, Cancer Patient Education, Supportive Care, AI-Based Counseling
Brief summary
This study aims to evaluate the impact of a ChatGPT-based digital counseling intervention on anxiety and depression in newly diagnosed cancer patients undergoing chemotherapy. Patients are randomly assigned to one of two groups: ChatGPT Group: Patients receive standard chemotherapy education plus access to ChatGPT, where they can ask up to 10 personalized questions about their treatment. Control Group: Patients receive only standard chemotherapy education provided by oncologists and nurses. The primary outcome is the change in anxiety and depression levels, measured using the Hospital Anxiety and Depression Scale (HADS), recorded at baseline and before the second chemotherapy cycle. Secondary outcomes include patient engagement patterns, the types of questions asked, and an assessment of ChatGPT's responses by oncologists. This study seeks to explore whether AI-based digital counseling can serve as a complementary support tool for newly diagnosed cancer patients.
Detailed description
Study Design This prospective, randomized controlled trial aims to evaluate the effects of a ChatGPT-based digital counseling intervention on anxiety and depression in newly diagnosed cancer patients undergoing chemotherapy. The study is conducted at two oncology centers between July 2024 and December 2024. Patients are randomly assigned (1:1) to either: ChatGPT-Assisted Counseling Arm: Patients receive standard chemotherapy education plus the ability to ask up to 10 personalized questions to ChatGPT. All AI-generated responses are reviewed for accuracy by oncology specialists. Standard Education Arm: Patients receive only clinician-led chemotherapy education, without AI-based support. Primary and Secondary Outcomes Primary Outcome: Change in Hospital Anxiety and Depression Scale (HADS-A, HADS-D) scores between baseline and prior to the second chemotherapy cycle. Secondary Outcomes: Differences in patient engagement behaviors and information-seeking patterns. Analysis of the accuracy and reliability of ChatGPT-generated responses. The role of AI in addressing sensitive or under-discussed topics in oncology education (e.g., sexual health, alternative medicine). Statistical Analysis The sample size is estimated based on detecting a 2-point difference in HADS scores (α=0.05, power=80%). Nonparametric tests (Mann-Whitney U test, Wilcoxon signed-rank test) are planned for within- and between-group comparisons. Multivariable logistic regression will assess independent predictors of anxiety reduction.
Interventions
BEHAVIORALChatGPT-Based Digital Counseling
Participants in this group received standard clinician-led chemotherapy education plus access to ChatGPT-based digital counseling. They could submit up to 10 personalized questions regarding their treatment, side effects, and supportive care. Oncology specialists reviewed ChatGPT-generated responses for accuracy. This intervention aimed to assess whether AI-assisted counseling could reduce anxiety and depression levels in newly diagnosed cancer patients.
BEHAVIORALStandard Clinician-Led Education
Participants in this group received only standard clinician-led chemotherapy education, delivered by oncology physicians and nurses. The education covered key aspects of chemotherapy, including potential side effects, treatment schedules, and supportive care recommendations. Unlike the ChatGPT group, these participants did not receive AI-assisted counseling or additional interactive educational tools.
Sponsors
Gazi University
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
SUPPORTIVE_CARE
Masking
SINGLE (Outcomes Assessor)
Masking description
This study used an outcomes assessor-blinded design, meaning that the researchers responsible for analyzing the primary and secondary outcomes were blinded to group allocation. However, participants, care providers, and investigators were aware of the assigned intervention.
Intervention model description
This study used a parallel assignment model in which participants were randomly assigned to one of two groups: (1) the ChatGPT-assisted counseling group, which received standard chemotherapy education plus access to ChatGPT for personalized inquiries, or (2) the control group, which received standard clinician-led chemotherapy education alone. Both groups were followed simultaneously to assess changes in anxiety and depression levels before the second chemotherapy cycle.
Eligibility
Sex/Gender
ALL
Age
18 Years to 85 Years
Healthy volunteers
No
Inclusion criteria
Age ≥18 years Newly diagnosed solid tumor (treatment-naïve) Scheduled to begin first-line chemotherapy ECOG performance status ≤2 Sufficient cognitive function to understand and complete the Hospital Anxiety and Depression Scale (HADS) questionnaire Ability to provide written informed consent
Exclusion criteria
Recurrent or metastatic cancer requiring urgent treatment Cognitive impairment that affects the ability to understand or complete the Hospital Anxiety and Depression Scale (HADS) questionnaire Severe psychiatric disorders (e.g., schizophrenia, bipolar disorder, severe major depressive disorder) that could interfere with study participation Concurrent participation in another interventional study related to psychological support or digital counseling
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Change in Anxiety and Depression Scores (HADS-A and HADS-D) | Time Frame: Change in Hospital Anxiety and Depression Scale (HADS-A, HADS-D) scores from baseline to prior to the second chemotherapy cycle (each cycle is 21 days). | The primary outcome is the change in Hospital Anxiety and Depression Scale (HADS-A, HADS-D) scores between baseline and the second chemotherapy cycle. The difference between groups will be assessed to determine the impact of ChatGPT-assisted counseling on anxiety and depression levels in newly diagnosed cancer patients. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Differences in Patient Engagement and Information-Seeking Behavior | Change in Hospital Anxiety and Depression Scale (HADS-A, HADS-D) scores from baseline to prior to the second chemotherapy cycle (each cycle is 21 days). | The number and type of questions asked by participants in the ChatGPT-assisted group will be analyzed to assess differences in information-seeking behavior compared to the standard education group. |
| Accuracy and Reliability of ChatGPT Responses | Baseline and Day 21 (prior to Cycle 2, each cycle is 21 days) | The oncology specialists will review all AI-generated responses and classify them as appropriate, inappropriate, or unreliable. The percentage of correct responses will be recorded. |
Countries
Turkey (Türkiye)
Outcome results
None listed