Artificial intelligence (AI) is increasingly recognized for its potential to transform medicine, providing benefits for physicians and patients across various aspects of care. Its integration into healthcare aims to personalize treatment, improve patient outcomes, and address challenges such as rising costs and resource limitations. AI applications already support treatment planning, disease diagnosis, outcome prediction, operational efficiency, and patient monitoring. However, adoption remains cautious within the medical community due to concerns about accuracy, unfamiliarity, and ethical implications.
AI comprises computer systems that learn from data to identify patterns and assist in decision-making. Machine learning (ML) handles structured data like lab results, while deep learning (DL) processes complex inputs such as medical images and clinical notes. Techniques like natural language processing (NLP) enable AI to interpret written language, and neural networks support image recognition and disease prediction. Studies have demonstrated AI’s diagnostic capabilities, including dermatology models performing on par with experts and echocardiographic analyses predicting cardiac conditions with high accuracy. Moreover, during the COVID-19 pandemic, AI facilitated outbreak detection and digital contact tracing through analysis of unstructured data.
Despite advances, a knowledge gap persists regarding AI’s role in holistic patient care, particularly in palliative care where quality of life is prioritized. Patients with blood cancers such as leukemia, lymphoma, and multiple myeloma face complex symptoms and disease courses, yet research on AI’s application in their palliative care remains limited.
This review introduces AI’s emerging role in palliative oncology, focusing on hematologic malignancies. Palliative oncology here refers to integrating symptom management, communication, and psychosocial support within oncology throughout the disease trajectory. Highlighting current capabilities, future promise, and areas requiring human judgment, the article aims to clarify practical AI applications for clinicians.
A systematic PubMed review was conducted, focusing on randomized controlled trials (RCTs) from 2023-2024 for AI in palliative care and RCTs from 2000-2024 for AI in hematologic malignancies. Relevant studies were selected based on applicability.
Recent clinical trials demonstrate AI’s potential in palliative care. AI decision support tools increased palliative care consultations and reduced readmissions. ML-triggered behavioral prompts improved serious illness discussions and curtailed aggressive therapies near end of life. NLP enhanced the accurate documentation of goals-of-care conversations, reducing manual workload. AI combined with wearable devices predicted short-term mortality in terminal cancer patients with high accuracy, underscoring the value of continuous physiological monitoring. Mobile patient-administered recordings improved informed decision-making and awareness of palliative care options. Collectively, these findings underscore AI’s role in optimizing patient care and clinical efficiency.
In symptom management, chronic pain—a major contributor to patient burden and healthcare costs—can benefit from AI’s pattern recognition capabilities. AI models, especially random forests, have effectively predicted pain intensity, opioid response, and chronic pain risk, outperforming traditional methods. Automated pain assessments using facial expressions, language, posture, and neurophysiological signals have standardized and improved accuracy. Clustering algorithms stratify patients by shared pain characteristics, guiding personalized treatments. AI has also been applied to other symptoms including fatigue, depression, anxiety, and gastrointestinal distress through analysis of electronic health records (EHRs) and patient-reported data. Wearable sensors provide real-time monitoring of vital signs and movement, correlating with survival outcomes and assisting prognostication and care planning.
AI supports decision-making in palliative care by predicting therapy responses from genomic data and clinical variables, assisting prognosis discussions, and optimizing drug dosing to reduce adverse effects. Algorithm-based predictions of patient preferences offer decision support when patients cannot express their wishes, though they require careful oversight to avoid bias from incomplete or subjective records. AI-powered telehealth tools extend care access in remote or underserved areas, facilitating virtual consultations and symptom monitoring.
In hematologic malignancies, AI applications span disease classification, prognosis, treatment optimization, and response assessment. Programs like LEAP in chronic myeloid leukemia guide personalized tyrosine kinase inhibitor therapy, improving survival. ML models aid relapse prediction, minimal residual disease detection, and response evaluation in acute myeloid leukemia and myelodysplastic syndromes. In lymphoma, DL and random forest models enhance diagnostic accuracy and risk stratification beyond conventional methods. Myeloma research uses machine and deep learning to predict survival and adverse events; AI-assisted PET/CT imaging quantifies bone marrow involvement, supporting treatment planning.
AI enhances symptom management in hematologic cancers by predicting complications such as anemia, infections, and treatment toxicities through EHR data and wearable devices. Noninvasive hemoglobin estimation via smartphone imaging and ML models predicting transfusion needs facilitate tailored interventions. AI-driven pain assessment tools inform medication strategies. Models also predict febrile neutropenia, sepsis, and respiratory distress, enabling early intervention. Psychological symptom detection through NLP analysis of notes and patient data supports timely mental health care, while AI chatbots provide ongoing symptom and medication support.
Personalized care is furthered by AI analytics differentiating patients by symptom burden and treatment response, guiding palliative interventions and therapy adjustments. AI-based monitoring tools track vital signs and adverse effects to enable proactive management. AI-driven decision support in transfusion medicine reduces unnecessary procedures while maintaining safety.
Accurate prognostication and risk stratification are critical in hematologic malignancies. AI integrates clinical, genomic, and imaging data, improving predictions of mortality, relapse, and treatment outcomes beyond traditional models. In stem cell transplantation, AI models predict graft-versus-host disease severity, transplant mortality, and relapse to tailor patient selection and management. AI also supports survivorship care by monitoring late treatment complications and optimizing follow-up schedules. At end of life, AI prognostic models identify candidates for early palliative care referrals, enhancing care quality.
Ethical considerations include data privacy, bias, consent, and preserving patient autonomy. AI should augment, not replace, clinical judgment and empathy in palliative care. Transparency, explainability, and continuous validation are essential to minimize disparities and avoid unintended care limitations. Clinician training, infrastructure support, and regulatory clarity are needed to ensure responsible AI adoption.
AI holds substantial promise to transform palliative oncology and hematologic cancer care by improving diagnostics, symptom management, prognostication, and personalized treatments. Future research should emphasize prospective studies and interdisciplinary collaboration to integrate AI effectively and ethically, ultimately enhancing patient-centered care across disease trajectories.
Leave a Reply