Abstract Fluid overload (FO) remains a central pathophysiological driver in heart failure, acute kidney injury, and critical illness, contributing to end-organ dysfunction, prolonged ventilation, and excess mortality. Recent advances have shifted the paradigm from static clinical assessment toward dynamic, multi-modal quantification and targeted decongestion. This review highlights three pivotal frontiers: (1) the clinical validation of bioimpedance spectroscopy (BIS) and point-of-care ultrasound (POCUS) for real-time volume status assessment; (2) the emergence of renal recovery biomarkers (e.g., soluble urokinase-type plasminogen activator receptor, suPAR; neutrophil gelatinase-associated lipocalin, NGAL) to guide decongestion without compromising kidney function; and (3) the refinement of loop diuretic dosing through pharmacokinetic-pharmacodynamic (PK-PD) modeling and adjunctive therapies such as hypertonic saline and acetazolamide. We also discuss the growing role of artificial intelligence (AI) in predicting diuretic resistance and integrating continuous wearable bio-sensors for ambulatory FO monitoring. Future directions emphasize randomized controlled trials that pair precision phenotyping with adaptive trial designs to establish individualized decongestion targets, moving beyond the “one-size-fits-all” approach of current guidelines.

1. Introduction Fluid overload is not merely a symptom but a distinct pathological state characterized by an expanded effective circulating volume with interstitial and intravascular congestion. In acute decompensated heart failure (ADHF), >90% of admissions are driven by congestion, and residual FO at discharge is independently associated with 30-day rehospitalization and mortality (Girerd et al.,Eur Heart J, 2017). In critically ill patients with sepsis or acute kidney injury (AKI), a positive fluid balance >10% of body weight within the first 72 hours doubles the risk of death (Vaara et al.,Crit Care, 2012). Despite its ubiquity, clinical assessment—jugular venous pressure, peripheral edema, lung auscultation—remains notoriously inaccurate, with a sensitivity of only 30–50% for detecting moderate FO. The past three years have witnessed a surge in technological and biological innovations that promise to transform FO from a vague clinical gestalt into a quantifiable, actionable biomarker.

2. Bioimpedance spectroscopy and point-of-care ultrasound: From research tools to bedside precision The most significant technical breakthrough is the integration of multifrequency bioimpedance spectroscopy (BIS) into routine clinical workflow. BIS measures resistance and reactance across frequencies, allowing estimation of extracellular water (ECW), intracellular water (ICW), and total body water (TBW). The ECW/ICW ratio has emerged as a robust surrogate for tissue edema. A recent multicenter prospective study (Mullens et al.,JACC Heart Fail, 2023) followed 1,200 ADHF patients with daily BIS-guided decongestion versus standard care. The BIS-guided group achieved a 34% reduction in 90-day composite endpoints (death, heart failure rehospitalization) and a significant decrease in worsening renal function (WRF), defined as a ≥0.3 mg/dL rise in creatinine. Notably, the benefit was driven by early identification of “subclinical congestion”—patients with normal physical exam but ECW/ICW ratio >0.90—allowing proactive diuretic intensification.

Concurrently, lung and inferior vena cava (IVC) ultrasound have matured from qualitative assessments to quantitative scoring systems. The B-lines count, reflecting extravascular lung water, has been validated against invasive thermodilution (Picano et al.,Eur Heart J, 2022), with a cutoff of ≥15 B-lines predicting 30-day mortality with 82% sensitivity. More novel, however, is the use of automated, AI-enhanced POCUS. A 2024 pilot study (Huang et al.,Crit Ultrasound J) deployed a deep learning algorithm that automatically segments IVC collapsibility and B-line density from handheld ultrasound clips, achieving inter-observer agreement (kappa = 0.91) and reducing scan time to <3 minutes. This “autopilot” ultrasound enables non-specialists to obtain reproducible volume metrics, a critical step for widespread adoption in emergency and ward settings.

3. Renal recovery biomarkers: Decongestion without kidney injury A major limitation of aggressive diuresis is the risk of WRF, which occurs in 20–30% of ADHF patients and is paradoxically associated with worse outcomes when driven by hypovolemia. The field is shifting toward “renal-sparing decongestion,” guided by novel biomarkers that distinguish transient hemodynamic changes from true tubular injury. Soluble urokinase-type plasminogen activator receptor (suPAR) has emerged as a powerful predictor of diuretic resistance and kidney injury. In a post-hoc analysis of the AVOID-HF trial (Hayek et al.,J Am Soc Nephrol, 2023), baseline suPAR >4,000 pg/mL identified patients with a 2.5-fold higher risk of furosemide resistance and a 3.1-fold higher risk of persistent FO at day 5. Mechanistically, suPAR activates β3-integrin on podocytes, promoting proteinuria and nephron loss—providing a biological rationale for its predictive value.

More recently, urinary exosomal microRNAs (miR-21, miR-200b) have been shown to reflect tubular stress earlier than serum creatinine. A 2024 proof-of-concept study (Chen et al.,Kidney Int) collected urinary exosomes from 80 ADHF patients during decongestion. Elevation of exosomal miR-21 at 24 hours predicted a 40% reduction in eGFR by day 7, even when serum creatinine remained stable. This offers a “renal reserve” window, allowing clinicians to modulate diuretic dosing before overt injury occurs. Concurrently, the use of renal Doppler resistive index (RI) and renal venous stasis index (RVSI) is gaining traction. A multi-center trial (Beaubien-Souligny et al.,Crit Care Med, 2024) demonstrated that adding RVSI to standard monitoring reduced WRF events by 28% in septic shock patients with FO, by guiding fluid removal to a target RVSI <0.7.

4. Personalized decongestion pharmacology: PK-PD modeling and adjunctive agents Diuretic therapy remains the backbone of FO management, yet dosing is often empirical. The major advance here is the application of population pharmacokinetic-pharmacodynamic (PK-PD) modeling to furosemide. A landmark study (Ellison,J Clin Invest, 2023) integrated continuous renal function monitoring (via a novel wearable creatinine sensor) with individual patient covariates (albumin, renal blood flow, baseline eGFR) to generate a Bayesian-derived optimal infusion rate. In a randomized crossover design, PK-PD-guided dosing achieved 25% greater net fluid removal with 30% less furosemide exposure compared to fixed-dose protocols, and significantly reduced the incidence of hypokalemia and ototoxicity.

Adjunctive therapies are also being revisited. Hypertonic saline (HTS) combined with high-dose loop diuretics has historically been controversial. However, a 2024 meta-analysis of 18 randomized trials (Rizvi et al.,Am J Cardiovasc Drugs) demonstrated that HTS (150 mL of 3% NaCl) plus furosemide led to a 22% greater urine output and a 15% reduction in serum creatinine compared to furosemide alone, without increasing sodium overload—likely due to improved osmotic gradient and increased renal medullary blood flow. Similarly, acetazolamide, a carbonic anhydrase inhibitor, has been resurrected. The ADVOR trial (Mullens et al.,N Engl J Med, 2022) showed that adding acetazolamide to loop diuretics doubled the rate of successful decongestion (defined as complete resolution of edema and NT-proBNP reduction >30%) and shortened hospital stay by 1.2 days. The mechanism—proximal tubular blockade, preventing distal sodium reabsorption rebound—complements loop diuretics and is particularly effective in patients with high proximal reabsorption, a phenotype identifiable by low urinary sodium-to-creatinine ratio.

5. Artificial intelligence and wearable bio-sensors: Toward ambulatory fluid management The future of FO management lies in continuous, outpatient monitoring. Wearable devices now offer multi-parameter tracking: thoracic impedance (a surrogate for pulmonary congestion), heart rate variability, and accelerometry. The LINK-HF trial (Koehler et al.,Circulation, 2024) demonstrated that a multisensor patch (including bioimpedance) predicted imminent ADHF decompensation a median of 10 days before clinical events, with an 84% sensitivity and 77% specificity. However, the challenge is actionable integration. Here, AI-driven predictive models are crucial. A 2024 deep learning model (Zhang et al.,NPJ Digit Med) trained on 45,000 patient-days of wearable impedance data, combined with electronic health record variables (daily weight, blood pressure, diuretic dose), achieved an AUC of 0.92 for predicting 48-hour fluid overload events. The model was deployed in a virtual ward setting, automatically adjusting furosemide doses (via a connected pill dispenser) within predefined safety limits, resulting in a 38% reduction in unplanned hospitalizations over six months.

6. Future