A Comfort-First Approach to Palliative Care: AI, Trends, and Home-Based Support
This article is based on my personal experience caring for my palliative mother, who at the time was expected to live less than two months. I am not a doctor. These reflections arise from witnessing how end-of-life care, despite good intentions, often brings unnecessary discomfort to patients even when death is near. True palliative care should focus on comfort, dignity, and relief from suffering—not on prolonging life through burdensome medical interventions.
In hospitals, routines designed for safety frequently go too far. Nurses and physicians must follow strict protocols requiring frequent vital checks, blood draws, and continuous monitoring. Much of this practice stems from fear of legal responsibility rather than genuine medical necessity. As a result, even patients in their final hours are often subjected to procedures that provide no benefit but cause distress. Many remain connected to machines until their last moments, while families watch helplessly as their loved ones endure pain in contradiction to the very essence of palliative care. What is needed is a systemic shift—from monitoring by default to comfort by default.
​
Artificial intelligence could help make this transformation possible. By analyzing non-invasive physiological data, AI can alert clinicians only when real attention is needed, reducing unnecessary tests and visits. When comfort-first principles are embedded in both policy and technology, patients can die peacefully while clinicians remain protected from liability.
A more humane approach lies in home-based, AI-assisted palliative care. Instead of relying on advanced, high-precision diagnostics that often require patient discomfort or hospital transfer, we should prioritize low-cost, comfortable, and timely assessments that capture meaningful physiological trends. Wearable sensors can measure heart rate and oxygen saturation, while bioimpedance or other non-contact methods can help track changes in fluid balance and muscle loss. Even if these measurements are inaccurate at a single point in time, their daily trajectories reveal valuable information about dehydration, edema, or gradual decline. What matters most is not precision at the moment of measurement, but the overall pattern of change that reflects the body’s ongoing state. A handheld ultrasound device, roughly the size of a smartphone, can also serve as a home-based diagnostic tool. For patients nearing the end of life, most essential diagnostics have already been completed, so ultrasound should function as a simple, non-invasive, and inexpensive supplement rather than a high-precision imaging modality. Its purpose is not to extend life through aggressive interventions but to support comfort, reduce anxiety, and provide clarity to families.
At the final stage of life, the emphasis should shift even more clearly toward recognizing trends rather than pursuing numerical precision. Highly simplified sensors are sufficient if they minimize distress. For example, HUINNO’s MEMO Patch and Daewoong’s MobiCare system offer wearable ECG solutions designed for greater comfort, using lightweight devices and simplified electrode configurations instead of traditional multi-lead setups. Invisible monitoring—such as under-mattress sensors, mmWave radar, and wrist or ring wearables—is preferable to avoid wires, skin irritation, and stigma. Intrusion should be kept to a minimum, with only one or two scheduled checks per day for blood pressure and spot ECG, while all other monitoring remains passive. The system is meant to fade into the background, preserving dignity and supporting uninterrupted family presence and peace.
​
I suggest developing an AI platform that integrates these diverse data streams, summarizes daily changes, and predicts deterioration. Doctors would receive concise updates and visit only when truly necessary, minimizing disruption to patients and families. Even low-resolution imaging can help detect conditions such as ascites or pleural effusion, guiding simple comfort-oriented procedures and offering reassurance through clear visual explanations. Initially, this model should focus on patients expected to live less than a year and no longer receiving curative treatment. With explicit family consent, AI monitoring would be used only to support comfort, not to prolong suffering. A shift toward home-based, AI-supported palliative care could restore the true spirit of medicine—helping people spend their final days in peace, comfort, and dignity.
​
I further suggest that such a system combine multiple forms of information, including physiological signals such as photoplethysmography (PPG), oxygen saturation (SpOâ‚‚), activity data, blood pressure, and spot ECG results, together with prior knowledge from CT or MRI reports, laboratory results, diagnoses, and medications. It would generate short-term deterioration and mortality risk, predict episodes of respiratory distress, and produce concise daily summaries for clinicians and families. To preserve comfort, the model would be trained with missing-modality dropout, ensuring reliability even when some measurements are skipped. Derived or cuffless signals would be interpreted primarily as trend indicators rather than absolute diagnostics. The system would also support bedside ultrasound as a low-burden adjunct, identifying major clinical problems such as pleural effusion, ascites, or bladder retention without requiring hospital transfer. By integrating all monitoring and imaging outputs into a unified telepalliative platform, clinicians could observe physiological data, imaging summaries, and laboratory trajectories on a single screen—enabling timely, compassionate, and minimally invasive care at home.
​
Using Bioimpedance and EIT Trends to Monitor End-of-Life Physiology
Bioimpedance may also offer a gentle way to monitor muscle and tissue changes near the end of life. In palliative care, what matters is not absolute accuracy but how electrical properties of tissue evolve over time. Devices similar to InBody products, which track trends in muscle mass, hydration, and fluid balance, could provide useful insight into gradual physiological decline with minimal discomfort. Electrical impedance tomography (EIT) may also have potential value, but its current form is too cumbersome for frail patients. To make EIT more practical, it would be necessary to create a user-friendly electrode belt that allows quick and simple attachment, and to develop deep learning tools that summarize a full day of EIT data into a short, meaningful video while automatically removing motion-corrupted segments. These steps would help reduce burden and make the technology compatible with comfort-focused care. For any such approach to be truly useful in palliative settings, the process must cause no stress, require minimal handling, and preserve patient rest. These ideas reflect my personal suggestions, and I do not know whether such methods are currently considered suitable in clinical practice. Nonetheless, time-based electrical trends—not exact numerical values—may serve as a low-burden way to recognize ongoing weakening and guide comfort-oriented care.
​
Low-Cost, Low-Dose, Low-Speed Rotational CBCT for Easy Maintenance in Hospice Care
In palliative and hospice settings, the primary goal of diagnostic imaging is clarity, not precision. Clarity supports decisions that prioritize patient comfort while avoiding unnecessary transport or invasive procedures. With this in mind, a low-cost (potentially less than 10% of the price of a conventional CT) and low-dose whole-body CBCT system—modeled after dental CBCT—could offer meaningful benefits despite its technological limitations.
Dental CBCT technology is widely available, relatively inexpensive, and delivers substantially lower radiation doses than standard medical CT. These systems typically have lower spatial resolution, slower gantry rotation speeds, and are more susceptible to artifacts, including those related to scatter and motion. Although these factors can reduce image accuracy, they are less problematic in palliative care, where the clinical goal is comfort-focused decision-making rather than high-fidelity diagnostic evaluation.
​
Implementing an ultra-low-cost CBCT system in hospice hospitals could greatly reduce the stress and logistical burden associated with transporting frail elderly patients to major medical centers for conventional CT. Instead of long wait times, ambulance transfers, and exposure to crowded emergency environments, a bedside or near-bedside CBCT scan could provide sufficient information for simple, compassionate interventions. Even low-resolution images may help clinicians determine whether symptoms stem from treatable mechanical problems—such as bladder retention, severe constipation, or large pleural effusions—or whether invasive procedures should be safely avoided.
​
The aim is not to replace advanced CT imaging, but to use ultra-low-cost CBCT as a practical, gentle, and ethically aligned tool that supports comfort-first care. When paired with AI-based image enhancement or automated image summarization, low-dose CBCT could evolve into a streamlined system that highlights clinically important findings without requiring extensive interpretation by already overburdened clinicians.
​
This concept reflects my personal suggestions and may not align with current clinical guidelines or regulatory standards. Nonetheless, the underlying principle remains: in palliative care, imaging should be simple, low-burden, and focused on peace rather than precision. Thoughtful, selective use of low-dose CBCT could contribute to a more humane end-of-life care model by reducing patient movement, avoiding unnecessary transfers, and supporting informed, compassionate decision-making.