Rik Tonino

For a large part based on the published article:
“Toward personalized transfusion strategies: The emerging role of
wearable biosensors in chronic anaemia management”

R.P.B. Tonino,1,2,3 M.R. Schipperus,4 and J.J. Zwaginga1,2
1 Leiden University Medical Center, Leiden, The Netherlands
2 TRIP Hemovigilance and Biovigilance Office, Leiden, The Netherlands
3 Groene Hart Ziekenhuis, Gouda, The Netherlands
4 Sanquin Blood Supply, Amsterdam, The Netherlands

Introduction
Throughout this thesis, we have tried to answer several pressing questions: what
are the effects of haemoglobin modulation? Is there a causal relationship between
Hb level and physiological or patient-reported outcomes? How can we measure
these outcomes effectively and accurately? And how should we interpret these
measurements to inform and optimize transfusion or donation strategies?

As Bram Stoker, author of Dracula, might have put it: “The blood is the life” – but how
much life is actually in the blood? Here we summarize our findings and reflect upon
their implications.

Chapter 2
Haemoglobin modulation affects physiology and patient-reported outcomes in
anaemic and non-anaemic subjects: An umbrella review

Background and context
Both low and high haemoglobin concentrations are known to affect physiological
performance and quality of life. Yet, despite decades of research, uncertainty remains
regarding how Hb modulation through interventions such as transfusion, phlebotomy,
iron supplementation or erythropoiesis-stimulating therapy, or even altitude
exposure, translates into measurable physiological and patient-reported outcomes.
Existing clinical thresholds for transfusion or target Hb levels often rely on arbitrary or
disease-specific conventions rather than on integrative physiological evidence.

To address this gap, we conducted an umbrella review synthesizing evidence from
published systematic reviews and meta-analyses that evaluated the relationship
between Hb modulation and both physiological and subjective outcomes in adults
with and without anaemia. The goal was to determine whether changes in Hb
concentration are consistently associated with improvements in measurable function
and perceived well-being across populations and interventions.

Study summary and findings
A systematic search across six databases up to April 2022 identified 33 systematic
reviews, comprising diverse populations (anaemic, non-anaemic, and polycythaemic)
and interventions aimed at modulating Hb levels. Eligible reviews reported at
least one physiological endpoint (e.g., VO2max, heart rate, exercise capacity) or a
patient-reported outcome (e.g., fatigue, dyspnoea, Quality of Life).

Quality assessment using AMSTAR-2 revealed that only 7 reviews were rated as high
quality, whereas 24 were judged critically low, highlighting substantial methodological
limitations within the available literature.

Across the included reviews, increases in Hb concentration were generally associated
with improvements in both physiological measures and patient-reported outcome
measures. These benefits were, as might be expected, most pronounced among
individuals with low baseline Hb levels, where modest Hb increases translated into
gains in functional capacity and perceived vitality. For example, in chronic kidney
disease, raising Hb toward 12 g/dL (7.5 mmol/L) was consistently linked to enhanced
exercise tolerance, improved fatigue and quality-of-life scores. Evidence regarding
the effects of supraphysiologic Hb increases or intentional Hb reduction (e.g.,
phlebotomy, haemodilution) was limited and heterogeneous.

Overall, we concluded that Hb modulation exerts measurable physiological and
patient-perceived effects, but the strength and consistency of evidence are constrained
by the quality and heterogeneity of existing studies.

Critical analysis
This umbrella review provides the first comprehensive synthesis of evidence linking Hb
modulation to physiological and subjective outcomes across a spectrum of populations
and interventions. We employed a transparent search strategy, adhered to PRISMA
guidelines, and used a formal quality appraisal tool. By combining data from multiple
disease contexts, the study successfully highlights overarching physiological patterns
that transcend individual disorders.

However, several limitations temper the interpretative value of its findings. The
predominance of low-quality reviews underscores the fragility of the evidence.
Considerable heterogeneity–in populations, interventions, endpoints, and analytic
methods–prevents reliable quantitative pooling or inference about causality.

Another limitation lies in the indirectness of umbrella-level evidence: overlapping
primary studies across reviews may have inflated perceived consistency, while
publication bias on expected results could favour the latter. Moreover, the review
focuses on associations rather than mechanistic pathways, leaving uncertain whether
observed improvements are directly mediated by Hb change or by concurrent
physiological and/or behavioural/psychological changes.

Despite these limitations, the review’s findings reinforce the clinical observation that
raising Hb from sub-physiologic to near-normal levels can enhance both functional
and subjective outcomes, whereas there is no evidence regarding Hb elevation above
physiologic norms.

Future studies should emphasize standardized physiological metrics, incorporate
digital or wearable monitoring for objective continuous assessment, and integrate
patient-reported outcome measures to better capture the multidimensional impact
of Hb modulation. Rigorous randomized trials and high-quality systematic reviews are
needed to clarify optimal targets and balance benefit against risk.

Chapter 3
Clinical practice for outpatients that are chronically red cell dependent: A survey in
the Netherlands

Background and context
For patients that are chronically transfusion-dependent, evidence guiding optimal
transfusion triggers, unit numbers and intervals in non-acute settings is scarce. In
the absence of robust data, clinical decisions regarding maintenance transfusion
regimens are often empirical and institution specific. We therefore designed a survey
to characterize current outpatient transfusion practices in the Netherlands, focusing
on three key dimensions: pre-transfusion haemoglobin thresholds, number of RBC
units per transfusion episode, and interval between transfusions. We further sought
to identify patient- and haematologist-specific factors influencing these decisions.

Study summary and findings
Between June and October 2020 we administered a web-based 44-question survey
to members of the Dutch Haematology Association. Fifty physicians responded
(response rate ~30%) representing both academic and community hospitals. The
responses revealed a very wide variation in transfusion strategies. Reported median
pre-transfusion Hb threshold was 8.0 g/dL (5.0 mmol/L), but individual thresholds
ranged from 6.4 to 9.6 g/dL (4.0–6.0 mmol/L). Most centres transfused 1–2 RBC units per
episode (range 0–3 units) at intervals of 2–4 weeks, though reported intervals ranged
from 1 to 12 weeks. Physicians reported that the clinical factors most strongly influencing
transfusion strategy were the presence of angina pectoris, cardiac failure, and dyspnoea.
Among subjective considerations, quality of life and patients’ self-sustainability were
regarded as most relevant. In contrast, fatigue, cognitive status, emotional state, and
vital parameters were considered to have less influence on transfusion decisions. We
concluded that while individualization may account for some variation in practice, the
wide heterogeneity observed underscores the necessity for randomized controlled
trials to inform evidence-based guidelines on the management of chronic transfu sion-dependent patients, including the identification of key clinical parameters.

Critical analysis
This survey addresses a significant gap in transfusion medicine by providing insights
into real-world practices for the management of chronically transfused outpatients.
Its strength lies in providing a quantitative snapshot of the diversity of practice across
Dutch haematologists and in highlighting the absence of consensus thresholds in a
routine care setting. The inclusion of both unit number and interval data, along with
threshold values, adds practical relevance for clinicians and guidelines.

Nevertheless, the study has several limitations. As a self-report survey, it reflects
reported practice rather than verified transfusion records, opening the possibility
of recall bias or social desirability bias. The relatively low response rate (30%) and
possible non-response bias may reduce generalisability. Importantly, the study does
not correlate transfusion strategies with clinical outcomes (such as symptom relief,
quality of life improvement or iron overload). Consequently, the variation observed
cannot be judged as beneficial, neutral or harmful. Differences in patient populations
(diagnosis, comorbidity, transfusion burden) across centres may explain much of the
heterogeneity yet cannot be delineated fully from the survey design.

Nonetheless, the findings have important implications. The documented variation
hints at potential under- or over-transfusion in some patient groups and underscores
the need for prospective comparative studies. From a clinical perspective, this survey
invites transfusion services to reflect on their local thresholds and intervals relative
to peer practice and to consider monitoring of patient outcome–both objective and
subjective–and resource utilisation. For guideline developers, the work emphasises
an urgent need to incorporate maintenance transfusion strategies into the broader
framework of patient blood management that mostly deals with perioperative
situations.

This survey of outpatient RBC transfusion practice in the Netherlands demonstrates
considerable heterogeneity in thresholds, unit numbers and intervals for chronically
transfused patients. While individualised care remains appropriate, the extent of
variation and lack of outcome data highlight a pressing need for prospective research
and evidence-based guidelines. Until such data emerge, clinicians are encouraged
to monitor not only Hb levels but also patient-centred outcomes to ensure optimal
practice.

Chapter 4
Remote patient monitoring in adults receiving transfusion or infusion for
haematological disorders using the VitalPatch® and AccelerateIQTM monitoring
system: Quantitative feasibility study

Background and context
Patients with hematologic disorders receiving treatments such as red blood cell
transfusions, chemotherapy, or immunotherapy are vulnerable to infusion-related
complications, often signalled by changes in vital signs. Traditionally, symptom
monitoring post-treatment is patient-reported, which can delay the detection of
adverse events. Wearable biosensors–already validated and used in cardiology
practice–in this line offer a solution. These devices enable continuous, real-time
monitoring of physiological parameters beyond the clinic or in less supervised (such as
home-) transfusion settings. This feasibility study explored whether integrating such
technology into outpatient care is acceptable and practical.

Study summary and findings
In this quantitative feasibility study, twelve adult patients with hematologic
conditions–including multiple myeloma, myelodysplastic syndrome, and non-Hodgkin
lymphoma–were enrolled to wear the VitalPatch® biosensor for a period of 12 days
around a planned infusion or transfusion. Patients received red blood cell transfusions
(n = 4), intravenous proteasome inhibitors (n = 4), or immunotherapy (n = 4). The
majority of participants (83%) were aged 60 years or older.

The VitalPatch® system–capable of measuring heart rate, respiration rate, skin
temperature, activity, heart rate variability and position–was generally well tolerated,
with minimal interference in daily activities. Patient-reported outcomes were collected
at baseline, day 1, day 5, and day 8 after application of the patch. Data indicated that
wearing the device had minimal impact on the patient's life. Ease of use, influence on
sleep, impact on follow-up of health, or discomfort with continuous monitoring did
not change between baseline and day 8. Patients reported low levels of discomfort,
although a 20% increase in skin irritation was reported, particularly when applying
a second patch. Both patients and nurses rated the wearability and usability of the
device positively. Nurses also reported that the sensor was easy to apply and provided
reliable measurements.

In this study we concluded that remote physiological monitoring using wearable
biosensors is feasible and acceptable among adult patients receiving outpatient
transfusions or infusions. These findings support the potential for broader
implementation of such technologies to improve patient safety and data collection in
haematology care.

Critical analysis
While this feasibility study highlights the potential of wearable biosensors like the
VitalPatch® to enhance outpatient monitoring in haematology, its limitations constrain
the broader implications of its findings. The small sample size (n = 12), short duration
of monitoring (12 days), and observational design restrict generalizability and preclude
conclusions about clinical efficacy or long-term outcomes.

The study shows that continuous monitoring is technically feasible and generally well
accepted by patients and healthcare staff. However, the absence of a control group
or direct linkage between recorded physiological changes and clinical actions limits
its ability to demonstrate tangible clinical benefit. Although not a goal of the study,
no adverse events or treatment modifications were reported in response to biosensor
data. Therefore, impact on clinical decision-making has yet to be formally evaluated.

Furthermore, while usability ratings were favourable, participant recruitment from a
single centre introduces potential selection bias. A more practical challenge is the
battery life of 4 days, while mild skin irritation in some patients also raises concerns
about long-term tolerability, especially in older adults with sensitive skin.

Broader implementation challenges–such as data interpretation, workflow integration,
and costs–remain unexplored. Nevertheless, the study does support continued
use and development of remote biosensor monitoring as part of a shift toward
personalized, proactive care in haematology.

Unpublished exploratory observations beyond the study scope
Although not part of the study’s predefined objectives, we conducted exploratory
analyses of the physiological data collected via the VitalPatch®. Notably, post-transfusion
measurements revealed patterns that appeared physiologically relevant and clinically
intriguing (Fig. 1). These preliminary findings prompted further interest in the potential
role of wearable biosensors in monitoring transfusion responses, ultimately serving as
the impetus for the subsequent pilot study.

Chapter 5
Remote monitoring of vital and activity parameters in chronic transfusion-dependent
patients: A feasibility pilot using wearable biosensors

Background and context
This study aimed to assess the feasibility of using wearable biosensors and web-based
assessments to monitor vital signs, activity levels, and cognitive function in transfusi on-dependent patients with myelodysplastic syndrome or thalassemia.

Study summary and findings
In this follow-up pilot study, five patients who regularly receive transfusions were
monitored starting 7 days before transfusion up until the following transfusion. The
monitoring tools included the AccelerateIQTM platform leveraging the VitalPatch®
biosensor, the Withings Steel HR smartwatch, and web-based cognitive and quality
of life assessments. Data collection was successful, with the AccelerateIQTM platform
yielding 97.8% high-quality usable data and the Withings Steel HR providing 98.9%
high-quality usable data. Post-transfusion assessments indicated a significant decrease
in heart rate and an improvement in sustained attention, suggesting a positive
physiological and cognitive response to transfusion. However, measures of physical
activity and self-reported quality of life did not show significant changes following
transfusion.

Critical analysis
This feasibility study demonstrates the potential of integrating wearable biosensors
and digital assessments into the care of transfusion-dependent patients. The high
quality and usability of the collected data indicate that such technologies can be
effectively employed in this patient population. The observed improvements in
heart rate and cognitive function post-transfusion provide preliminary evidence that
objective physiological and cognitive metrics may serve as valuable indicators of
transfusion efficacy.

Besides the small sample size, the lack of significant changes in physical activity
and quality of life measures may be due to the short–one single transfusion cycle–
monitoring period. Moreover, patient-related factors such as age, frailty, comorbidities,
and individual circumstances substantially affect these parameters.

Moreover, reliance on heart rate alone is likely insufficient to support algorithm-driven
decision-making in transfusion care. More nuanced physiological markers such
as heart rate variability, or yet-to-be-identified proxies of transfusion need–like a
gradually declining condition–will likely be necessary to develop robust, clinically
relevant algorithms.

Future research should explore the integration of wearable biosensor data into clinical
decision-making processes. Such studies could assess whether continuous monitoring
can lead to more personalized transfusion schedules, improved patient outcomes, and
enhanced quality of life. Furthermore, evaluating the cost-effectiveness and patient
adherence to long-term use of wearable devices will be essential for widespread
implementation.

Chapter 6
Heart rate changes after phlebotomy in polycythaemia vera and healthy donors:
An observational case-crossover pilot study

Background and context
While low Hb is linked to fatigue and reduced quality of life, excessive levels–as seen in
polycythaemia vera (PV)–can increase blood viscosity, potentially also compromising
oxygen delivery. Despite this, the physiological effects of modest Hb reduction in
non-anaemic individuals or in PV patients undergoing phlebotomy remain poorly
studied. This pilot aimed to explore whether small decreases in Hb levels elicit
measurable physiological responses, and whether consumer-grade biosensors can
detect them.

Study summary and findings
The study involved six healthy blood donors and five PV patients who were regularly
monitored via smartwatches for one week before and after phlebotomy. Among
healthy donors, a significant increase in heart rate post-phlebotomy was observed,
particularly during daytime hours, with gradual normalization. PV patients showed
no consistent change in heart rate. Activity levels rose briefly in healthy donors but
declined post-phlebotomy in PV patients, with modest recovery. Quality-of-life scores
showed minimal change in both groups.

Critical analysis
The study effectively illustrates that even small Hb reductions can provoke measurable
physiological changes in healthy individuals, notably increased heart rate. However,
the small, homogeneous sample–especially the all-female healthy donor group–limits
generalizability. The short follow-up period may not capture long-term changes, and
the observed variations in step count and quality-of-life scores may reflect natural
fluctuations more than true physiological shifts. The consistently high (100%) quality
of life score in the healthy donor group indicates a limitation of the EQ-5D-5L
questionnaire, where the ceiling effect prevents variation in responses.

Notably, PV patients exhibited greater physiological stability, suggesting that either
a slightly higher baseline Hb might reduce the impact of phlebotomy, or possibly
adaptation to chronic phlebotomy–no newly diagnosed PV patients with much
higher Hb values were investigated. The study underscores the potential of wearable
devices for passive physiological monitoring in diverse clinical settings. The findings
suggest that even subtle changes in normal haemoglobin levels can be detected by a
smartwatch, making it an intriguing technique. However, given the very small sample
size, this pilot study should be viewed as an initial signal pointing toward the potential
for larger-scale investigations.

Chapter 7
Heart rate changes in chronically red cell transfusion-dependent patients: a dose dependent effect of red cell transfusion–A randomized crossover trial, interim analysis

Background and context
As a follow-up to earlier pilot studies, we are currently conducting the REMOTE-2
study–a multicentre, randomized, within-subject crossover trial in outpatient clinics.
The study investigates the physiological and patient-reported effects of varying red
blood cell transfusion strategies in adults with chronic transfusion dependence due
to myelodysplastic syndromes, myeloproliferative neoplasms, or overlap conditions.
Each participant receives their standard-of-care transfusion regimen, followed by
two intensified regimens: one with one additional red blood cell unit (SoC+1), and
one with two additional units (SoC+2). The primary objective is to assess the effect
of increasing transfusion-modulated Hb on heart rate using wearable biosensors.
Secondary objectives include evaluating changes in physical activity, quality of life,
and cognitive function. Heart rate and activity were continuously monitored with a
wrist-worn biosensor; QoL, fatigue, and cognition were assessed at predefined time
points before and after transfusion. Due to slow enrolment, an interim analysis of the
first 12 participants was undertaken to provide early insights and determine whether
continuation of the study remained justified.

Summary and results
An increase in haemoglobin level was followed by a dose-dependent reduction in heart
rate. SoC transfusions yielded a small non-significant decrease, whereas SoC+1 and
SoC+2 resulted in progressively larger reductions. Heart rate reductions persisted into
the late post-transfusion phase following SoC+1 and SoC+2. Daily step count showed
wide inter- and intra-individual variability with no consistent differences between
regimens, and no temporal pattern suggestive of transfusion-related improvement.

Nor were any statistically significant cognitive effects of transfusion volume observed.
However, descriptive patterns suggested small, inconsistent improvements in
planning, working memory strategy, and sustained attention–most apparent in the
SoC+2 regimen. Whether these reflect transfusion effects or practice/familiarity
effects is uncertain.

QoL and fatigue scores did not show statistically significant improvements.
Nevertheless, trends were compatible with short-term benefit: reduced physical
burden and fatigue at the early post-transfusion visit following SoC+1 and SoC+2,
with some persistence of improvement in the SoC+2 regimen. These findings mirror
earlier small trials suggesting better symptom control under more liberal transfusion
strategies.[1,2]

Overall, the interim data suggest that higher transfusion volumes reduce heart rate in
a dose-dependent manner, with non-significant but clinically plausible trends toward
improved patient-reported outcomes.

Critical analysis
This interim analysis provides an early contribution to a field with limited empirical
evidence, addressing the clinically relevant question of whether higher haemoglobin
targets in chronically transfusion-dependent patients yield measurable physiological
or patient-reported benefits.

The most compelling finding is a dose-dependent reduction in heart rate with
higher transfusion volumes, supporting the hypothesis that increased haemoglobin
decreases cardiopulmonary compensation and enhances physiological reserve. This
effect remained significant after conservative statistical correction, reinforcing its
validity and demonstrating the sensitivity of wearable biosensors to detect subtle,
clinically meaningful changes.

Interpretation is constrained by the small sample size, which limits power for
secondary outcomes and may obscure effects on quality of life, cognition, or activity.
High variability in step counts and self-reported measures further reduces sensitivity,
while cognitive results may be confounded by practice effects, and remote CANTAB
testing may be less reliable in older populations.[3] Additionally, adjustments to
SoC transfusion volumes to avoid excessive unit numbers introduce complexity in
interpreting comparisons. For patients with a three-unit SoC–where a SoC + 2 strategy
would result in five units and increase the risk of transfusion-associated circulatory
overload–the SoC dose was reduced to two units, yielding SoC, SoC + 1, and
SoC + 2 regimens of two, three, and four units. This also shifts the SoC comparator in
this crossover design, as it no longer matches patients’ usual baseline dose.

The absence so far of clear improvements in activity or QoL does not preclude
clinical benefit, as chronic anaemia may require sustained haemoglobin elevation for
functional gains on physical or mental fitness to emerge, and current instruments may
lack sensitivity to short-term changes. Minimizing haemoglobin fluctuations, rather
than achieving transient post-transfusion peaks–may ultimately prove more relevant
to patient well-being. More frequent transfusions preferably at home in this respect
should be investigated with wearables additionally monitoring potential transfusion
reactions/side effects.

Overall, this interim analysis provides preliminary evidence supporting the biological
plausibility of benefit from more liberal transfusion strategies in chronic anaemia. Higher
transfusion volumes consistently reduced heart rate in chronically anaemic, transfu sion-dependent patients, suggesting heart rate may serve as a useful intermediate
marker of transfusion effect, with early signals also hinting at potential improvements
in symptoms and cognitive function. The study further demonstrates the feasibility of
integrating wearable technology into transfusion trials. While these findings require
confirmation in larger, fully powered cohorts, they support continued investigation of
individualized, outcome-oriented transfusion strategies beyond traditional restrictive
thresholds, including evaluation of whether sustained Hb elevation translates into
meaningful gains in function, cognition, or quality of life and how such benefits
balance against risks such as iron overload.

Other currently running trials
Another transfusion study incorporating biosensor technology is REDDS: Study 2.[4]
This investigation compares patients’ standard transfusion regimens with a fixed weekly
transfusion schedule to investigate the effect of reduced Hb fluctuations. In addition
to assessing health-related quality of life (HRQoL), the study employs accelerometry
to objectively measure patient activity levels throughout the study period. Lastly, the
OURA feasibility pilot study, using a finger ring based biosensor, will assess whether
patients with myelodysplastic syndromes (MDS) can maintain Oura ring use for more
than 70% of the time over a three-month period. Secondary exploratory analyses
will examine correlations between wearable-derived physiological parameters and
transfusion dates and self-rated physical functioning.[5]

Current applications of biosensors in other fields
While a focused approach is essential to advancing transfusion-related biosensor
development, insights from other fields can inform and accelerate progress. A
leading example is the use of hybrid closed-loop systems in diabetes management,[6]
which integrate continuous glucose monitoring with algorithm-driven insulin delivery
and remote data sharing, exemplifying the potential of biosensor-guided therapy.

This model has also spurred interest in wearable biosensors for neuropsychiatric
applications, including the detection of stress-related biochemical markers.
[7] In wound care, biosensors are employed to monitor parameters such as pH,
temperature, moisture, and inflammatory biomarkers, enabling timely intervention
and improved healing outcomes.[8] Similarly, cardiovascular research utilizes
biosensors in flexible comfortable formats (patches textiles, contact lenses,
tattoos) to continuously track vital heart-related signals, moving beyond simple
measurements such as heart rate, toward integrated cardiovascular health
tools–an approach that may also inform transfusion monitoring strategies.[9]

General discussion
Transfusion research has traditionally focused on broad clinical outcomes such
as survival and hospital length of stay–primarily in patient groups with acute
anaemia–leading to the conclusion that a restrictive transfusion strategy is generally
preferable–an approach informed by the scarcity and cost of blood products as well
as the undeniable, sometimes severe, adverse effects associated with transfusion. Yet
for patients with chronic, transfusion-dependent anaemia, quality of life and objective
measures of functional well-being should be integral considerations. These patients’
physical and psychological states are shaped by highly individual and timing-specific
factors, and emerging technologies such as wearable devices may offer continuous,
patient-centred assessment that could support more personalized transfusion care.

Despite decades of research, uncertainty remains regarding how Hb modulation
translates into measurable physiological and patient-reported outcomes. In Chapter
2, where we evaluated all available evidence from reviews, we concluded that Hb
modulation exerts measurable physiological and patient-perceived effects, but the
strength and consistency of evidence are constrained by the quality and heterogeneity
of existing studies.[10]

In clinical practice, this uncertainty is reflected in the substantial variation in transfusion
thresholds, unit numbers, and intervals reported in the Netherlands, underscoring
the need for prospective evidence to guide more standardized and outcome-driven
transfusion policies.[11]

Despite the well-documented impact of anaemia on functional and cognitive decline,
mobility, and overall physical performance,[12–15] especially in elderly populations,
a causality behind these associations is not proven; possibly, lower haemoglobin
concentrations may simply indicate a more advanced disease, with consequently
poorer QoL scores independent of the anaemia.

These unresolved uncertainties underscore the need for alternative markers to
evaluate transfusion effects and support more individualized decisions. One approach
explored in the present work involved the use of wearable biosensors and web-based
assessments.

The feasibility studies we performed indicate that biosensors are well tolerated and
technically viable in outpatient settings, including among older adults.[16,17] Building
on this foundation, our interim analysis showed that transfusing more RBC units was
associated with a dose-dependent reduction in heart rate with the suggestion of a
sustained effect up until the next transfusion.[18] The observed heart rate reductions
likely reflect an alleviation of the compensatory tachycardia characteristic of chronic
anaemia, thereby improving physiological reserves.

These physiological changes have not yet been shown to translate into significant
improvements in secondary outcomes such as quality of life, fatigue, activity or
cognitive function; an expected limitation given the sample size, though several trends
were suggestive. Trends towards improved cognitive performance may reflect either
subtle transfusion-related effects or practice-related improvement, especially given
the challenges older adults may face with web-based assessments. Similarly, trends
toward reduced fatigue and improved physical quality-of-life domains align with the
hypothesis that higher Hb levels enhance oxygen delivery and reduce compensatory
strain, consistent with findings from liberal versus restrictive transfusion trials.
[1,2] These findings, however, should be considered as hypothesis-generating and
should be interpreted cautiously given the small sample size and variability in patient
responses.

In contrast, daily step counts did not reflect transfusion effects and appear driven
more by individual lifestyle patterns than by Hb levels, indicating that step counts are
unlikely to serve as reliable functional markers. More structured assessments, such
as the 6-minute walk test, and long-term monitoring periods may more accurately
capture transfusion-related changes in functional capacity and a gradually changing
lifestyle.

Given the multifactorial determinants of quality of life, it remains uncertain to what
extent the increase in haemoglobin level alone produces clinically meaningful
improvements. It is plausible that reducing the amplitude of haemoglobin fluctuations,
rather than simply increasing Hb peaks, may yield greater benefits, a hypothesis
warranting further investigation. The latter is the subject of the presently running
REDDS2 study.

While these early findings are encouraging, they require cautious interpretation.
Key uncertainties persist regarding the reliability and interpretation of biosensor
signals, recommended clinical responses to alerts, and the downstream impact of
biosensor-guided transfusion strategies on mobility, QoL, healthcare use, and survival.
Artificial intelligence may enhance signal fidelity by distinguishing anaemia-related
physiological changes from day-to-day variability in patients’ activities and
comorbidities. Ultimately, large randomized trials with longer observation periods
comparing biosensor-guided strategies with current Hb-threshold-based practice
are needed to determine their true clinical value, incorporating both short-term
physiological outcomes and long-term patient-centred endpoints.

Before biosensor-supported transfusion management can be implemented at scale,
several practical and methodological challenges must be addressed. A key priority
is establishing validated cut-off points and alert thresholds, as the current analyses
demonstrate that biosensors can reliably detect transfusion-related physiological
changes but do not yet define when these changes should trigger clinical action. Without
such criteria, alerts may accumulate unnecessarily, increasing the likelihood of false
positives and contributing to alert fatigue, with the risk that clinically relevant signals
are overlooked. Additional considerations include integration with electronic health
records, workflow implications for clinicians, and cost-effectiveness. Nevertheless,
in an aging population with increasing healthcare demands, biosensor-guided
transfusion approaches hold considerable promise for personalizing care, optimizing
resource use, and ultimately improving patient-centred outcomes.

Future applications of biosensors in transfusion medicine
We anticipate that wearable biosensors will become increasingly integrated into
transfusion medicine, particularly in the context of the ongoing shift toward
decentralized and patient-centred models of care. The convergence of continuous
remote monitoring via cloud-connected wearable devices with personalized,
machine-learning–driven analytics holds considerable promise for the early detection
of anaemia-related physiological deterioration. The anticipated development
of multimodal biosensors that can combine electric signals from the heart with
biomarkers from blood or tears could substantially reduce the reliance on frequent,
routine outpatient visits for haemoglobin monitoring, thereby alleviating the workload
of healthcare professionals.

In parallel, the development of patient-facing smartphone applications may provide
opportunities for disease education, therapy guidance, and even advanced care
planning, further supporting patient autonomy while decreasing the time and
resources required from clinical staff. Importantly, however, attention must also be
given to equitable access. Without deliberate safeguards, these technologies may
disproportionately benefit patients with greater resources, digital literacy, or access
to robust healthcare systems, thereby widening existing disparities; considerations of
affordability, usability, and inclusivity in algorithm design will therefore be essential.

Considerations for future trials
Definitive evaluation of biosensor-guided transfusion strategies will ultimately
require a large international, multicentre trial. A pragmatic design could compare
current standard-of-care transfusion practices with an intervention arm in which
transfusion decisions are supported by biosensor-derived algorithms enhanced
by machine-learning techniques. Key outcomes should include total transfusion
requirements, healthcare utilization and associated costs, physiological responses
captured by biosensors (for example, whether algorithm-guided patients achieve
more sustained reductions in heart rate), patient-reported quality-of-life measures,
and longer-term functional outcomes such as the 6-minute walk test, grip strength,
and acute-care presentations.

However, the feasibility of such a trial is constrained by the limited pool of eligible
patients. As demonstrated by the slow recruitment in our own study, achieving
adequate sample sizes would require substantial resources and extended timelines.
Exploring analogous populations–such as athletes or patients with heart failure–may
offer insights into which physiological parameters best reflect functional status and
could help refine outcome selection for future transfusion trials.

Careful consideration must also be given to the choice of biosensor. With rapid
technological turnover, devices selected at trial initiation risk becoming obsolete
by completion. Ideally, the biosensor should be wireless, capable of autonomous
cloud-based data transfer, and require minimal patient involvement to ensure
consistent data capture. Questionnaires and cognitive assessments, if included, should
be concise and selected with care, as anaemia-related fatigue in this frail population
may influence performance unless this effect forms part of the research question.

Conclusion
Taken together, the work presented in this thesis underscores the longstanding
uncertainty surrounding how haemoglobin modulation translates into meaningful
physiological and patient-reported outcomes, and how this uncertainty continues to
shape the wide variation observed in current transfusion practice. Across our studies,
we found that haemoglobin augmentation consistently reduces compensatory
tachycardia, suggesting a tangible improvement in physiological reserve. However,
effects on broader functional outcomes, cognition, activity, and quality of life remain
less clear, limited both by small sample sizes and the inherent complexity of these
multidimensional endpoints.

Our feasibility studies demonstrate that wearable biosensors and web-based
assessments are well tolerated, technically robust, and capable of capturing
subtle physiological changes in chronically transfused patients. Yet their ultimate
clinical value–the extent to which biosensor-derived signals can guide transfusion
decisions, predict deterioration, or improve patient-centred outcomes–remains to
be established. Addressing these questions will require larger, rigorously designed
trials that evaluate not only short-term physiological responses but also long-term
impacts on functional capacity, quality of life, iron overload, healthcare utilization,
and survival.

Despite the challenges that lie ahead, the integration of continuous monitoring with
advanced analytics offers a promising path toward more individualized, responsive,
and outcome-driven transfusion care. As healthcare systems adapt to an aging
and increasingly complex patient population, the development of physiologically
informed, patient-centred transfusion strategies has the potential to substantially
enhance both clinical outcomes and quality of life for those living with chronic
anaemia.

To borrow once more from Bram Stoker – “The blood is the life”:
The challenge ahead is to ensure that every unit transfused not only sustains life,
but meaningfully enriches it.