Over the last four decades, advanced diagnostic imaging modalities such as CT and MRI have achieved remarkable progress, providing unprecedented precision and enabling earlier detection of disease. However, their high cost, demanding maintenance requirements, and reliance on specialized personnel make them unaffordable for many smaller clinical settings, especially rural hospitals and hospice facilities. For elderly, frail, or terminally ill patients living far from major medical centers, undergoing these imaging tests more than once often offers limited clinical benefit relative to the travel burden, financial cost, and physical discomfort involved.

​

When follow-up imaging is needed to observe changes over time, it must be performed in a nearby, easily accessible location. This, in turn, requires low-maintenance imaging systems that can be operated by non-specialists. A next-generation approach is needed—one that prioritizes affordability, reliability, and simplicity while supporting convenience-focused clinical decision-making. I believe research should focus on developing simplified full-body CBCT, low-field MRI, and AI-assisted handheld ultrasound systems that could be produced at roughly 10% of the cost of a conventional CT scanner.

 

Although these low-cost devices will inevitably produce lower-quality signals, artificial intelligence can partially compensate for these limitations through advanced reconstruction, denoising, and artifact reduction. While these ultra-low-cost systems don't achieve high levels of diagnostic performance, when combined with AI, they can provide essential information at a much lower cost and significantly improve accessibility for patients with mobility issues or in underserved areas.

Bioimpedance may offer a gentle and practical way to observe clinically meaningful physiological trends near the end of life, especially when combined with AI-based signal analysis that can extract subtle patterns from noisy, imperfect data. In palliative settings, the priority is not diagnostic precision but the quiet recognition of gradual changes in muscle integrity, hydration balance, and organ reserve. Because electrical impedance reflects intracellular and extracellular fluid distribution, muscle conductivity, and overall tissue composition, it naturally captures processes associated with weakening, reduced intake, and systemic stress. When these measurements are acquired repeatedly, AI tools can help interpret day-to-day fluctuations, smooth motion artifacts, and highlight the underlying trajectory of decline without requiring invasive testing.

​

As mobility and oral intake decrease, skeletal muscle impedance trends downward, and this slow change can serve as a nonintrusive indicator of diminishing strength. Bioimpedance is also sensitive to hydration shifts; gradual increases in extracellular fluid or signs of dehydration may correspond to worsening cardiac, renal, or metabolic function. Measures such as phase angle, while not precise, often correlate with cell health and frailty. AI-driven time-series models could track these variables more robustly, distinguish meaningful trends from incidental noise, and provide caregivers with an interpretable summary of the patient’s physiological trajectory. Because the goal in palliative monitoring is not to diagnose but to understand the direction and rate of change, AI-assisted trend analysis may help clinicians and family members recognize ongoing decline with very little burden on the patient.

​

Electrical impedance tomography (EIT) may also have a role, particularly for monitoring thoracic physiology, but in its current form it remains too cumbersome for frail individuals. To make EIT feasible, it would be necessary to develop a comfortable and easy-to-attach electrode belt capable of stable, low-stress monitoring. Deep learning tools would then be needed to compress an entire day of EIT measurements into a short, clinically meaningful video that automatically removes motion-corrupted segments and highlights ventilation or fluid-related patterns. Such automation would allow EIT to function as a background monitoring tool rather than a technical procedure, making it more compatible with comfort-focused care.

​

Ultimately, for bioimpedance, EIT, or any AI-enabled method to be meaningful in palliative settings, the entire process must avoid causing stress, require minimal handling, and preserve patient rest. The ideas presented here are personal suggestions rather than established clinical guidelines. Nonetheless, time-based impedance trends—interpreted and stabilized through AI-assisted analysis—may provide a realistic, low-burden way to acknowledge ongoing physiological weakening and support thoughtful, comfort-oriented decision-making in the final stage of life.