Page 131 - EJMO-9-1

Eurasian Journal of Medicine and
Oncology
Oncology care with AI chatbots and assistants

cancer), treatment phase (chemotherapy), prior In an oncology chatbot system, a patient experiencing
symptoms, and any previously logged interactions nausea might receive personalized advice tailored to their
with the assistant. needs. For example, the chatbot could suggest symptom
The patient has been diagnosed with breast cancer management strategies, such as anti-nausea medications,
and is currently undergoing chemotherapy. Previous dietary changes, or relaxation techniques, based on the
interactions have highlighted that common side patient’s treatment history and preferences. It would
effects, such as nausea and fatigue, have been reported. also proactively offer reminders to stay hydrated, which
iii. Step 3: NLP for intent detection are crucial for managing nausea, and schedule potential
NLP: The AI assistant identifies the main intent as follow-up interactions to assess the patient’s condition. The
“symptom management” for “nausea” and detects a response style would be empathetic and supportive, offering
secondary emotional need (patients may be distressed). comfort and understanding while providing additional
iv. Step 4: Personalized response tailoring resources, such as links to articles on managing nausea
The assistant analyzes the patient’s symptoms and during cancer treatment. Importantly, the system would
matches nausea-related advice to their chemotherapy ensure that all interactions are encrypted and comply with
context, ensuring the response is tailored specifically data security standards, such as HIPAA, to maintain patient
to chemotherapy-induced nausea. The assistant then confidentiality. For a straightforward case, such as nausea,
provides response options, including dietary tips, human escalation would not be required, as the system’s
breathing exercises, and over-the-counter remedies, advice and support would sufficiently address the patient’s
to help manage the symptoms effectively. needs. This approach aligns with current developments in
v. Step 5: Generate a personalized response AI-driven healthcare tools, which are increasingly being
Assistant response: “I’m here to help. Nausea can be designed to offer personalized, empathetic, and secure
common during chemotherapy. Here are some tips patient interactions while minimizing the need for human
that might help: Try eating smaller, more frequent intervention in routine cases.
meals, focus on bland foods, and stay hydrated by 3.4. Data and evidence
sipping water throughout the day. Ginger tea or
peppermint tea can also help ease nausea. If symptoms AI technologies have shown significant promise in
persist, you can reach out to your care team to discuss improving clinical outcomes in oncology, particularly
other anti-nausea options. Would you like reminders through applications, such as early cancer detection,
to stay hydrated or for a follow-up?” precision medicine, and personalized treatment planning.
vi. Step 6: Proactive support and adherence Deep learning models and AI-driven imaging analysis have
Proactive offers: The assistant offers to send hydration enhanced diagnostic accuracy, with studies demonstrating
reminders or follow-up prompts to check in on nausea dermatologist-level accuracy in skin cancer detection using
22
6
symptoms. CNNs. Similarly, Liu et al. demonstrated AI’s ability
vii. Step 7: Is remote support required? to predict malignancies from mammography scans with
Remote care check: Since the patient is not in a remote high sensitivity and specificity, outperforming traditional
area, no additional telemedicine support is needed. image analysis methods. AI tools have also revolutionized
viii. Step 8: Continuous learning and medical updates treatment planning by predicting chemotherapy resistance
9,17
Learning integration: The interaction is logged for and optimizing radiotherapy outcomes. In addition,
20
continuous improvement, helping the assistant better AI-driven genomic analysis has contributed to identifying
recognize similar cases and provide updated support. mutations associated with poor survival rates, allowing
ix. Step 9: Data security and privacy early intervention and targeted therapies. A meta-analysis
24
Security check: The assistant encrypts all data shared by Tan et al. further highlights that ML models improve
and ensures it complies with healthcare privacy survival predictions by 15% compared to traditional
regulations such as HIPAA. clinical approaches.
x. Step 10: Need human intervention? AI-powered virtual assistants and chatbots have
Complexity assessment: Since the patient’s needs were transformed patient engagement by offering personalized,
straightforward, no human intervention was required. accessible support. Studies have shown that AI tools
xi. Step 11: Final response to patient reduce patient stress and anxiety through real-time
Response to patient: The assistant confirms, “I’ve emotional support and sentiment analysis. In addition,
5,7
noted your preference, and you will receive hydration AI-driven reminders and adherence monitoring have
reminders twice daily. Feel free to reach out if you improved treatment adherence, with Lee et al. reporting
17
need further support.” a 20% increase in medication adherence due to AI-driven

Volume 9 Issue 1 (2025) 123 doi: 10.36922/ejmo.6251