Health State Classification Research Articles

Intelligent fault diagnosis for rotating machinery using deep Q-network based health state classification: A deep reinforcement learning approach

Fault diagnosis methods for rotating machinery have always been a hot research topic, and artificial intelligence-based approaches have attracted increasing attention from both researchers and engineers. Among those related studies and methods, artificial neural networks, especially deep learning-based methods, are widely used to extract fault features or classify fault features obtained by other signal processing techniques. Although such methods could solve the fault diagnosis problems of rotating machinery, there are still two deficiencies. (1) Unable to establish direct linear or non-linear mapping between raw data and the corresponding fault modes, the performance of such fault diagnosis methods highly depends on the quality of the extracted features. (2) The optimization of neural network architecture and parameters, especially for deep neural networks, requires considerable manual modification and expert experience, which limits the applicability and generalization of such methods. As a remarkable breakthrough in artificial intelligence, AlphaGo, a representative achievement of deep reinforcement learning, provides inspiration and direction for the aforementioned shortcomings. Combining the advantages of deep learning and reinforcement learning, deep reinforcement learning is able to build an end-to-end fault diagnosis architecture that can directly map raw fault data to the corresponding fault modes. Thus, based on deep reinforcement learning, a novel intelligent diagnosis method is proposed that is able to overcome the shortcomings of the aforementioned diagnosis methods. Validation tests of the proposed method are carried out using datasets of two types of rotating machinery, rolling bearings and hydraulic pumps, which contain a large number of measured raw vibration signals under different health states and working conditions. The diagnosis results show that the proposed method is able to obtain intelligent fault diagnosis agents that can mine the relationships between the raw vibration signals and fault modes autonomously and effectively. Considering that the learning process of the proposed method depends only on the replayed memories of the agent and the overall rewards, which represent much weaker feedback than that obtained by the supervised learning-based method, the proposed method is promising in establishing a general fault diagnosis architecture for rotating machinery.

Confirmatory analysis of a health state classification system for people living with dementia: a qualitative approach.

Objectives This paper aims to confirm the content validity of the domains identified during the development of the Alzheimer’s disease – five dimensions (AD-5D) algorithm for the quality of life – Alzheimer’s disease (QOL-AD) and to identify the rationale for stated quality of life preferences. Methods Focus groups were conducted to elicit the priorities for quality of life in dementia from three perspectives: the person with dementia; family caregivers; and the community. Participants were recruited through industry research partners (long-term care providers) based on knowledge of their experience with dementia. Three focus groups were conducted – one each in Brisbane, Sydney and Adelaide, Australia – between November 2016 and February 2017. Each focus group included participants providing a different perspective on dementia – people with dementia ( n = 3), caregivers ( n = 9) and general community members or relatives of residents of a long-term care facility ( n = 10), although some groups contained one participant with a different perspective. The focus groups were used to validate the AD-5D domains and examine quality of life preferences across the three perspectives. Thematic analysis was used to identify the priorities underlying preference selection. Results All activities affecting the quality of life for people with dementia could be mapped to one of the five AD-5D domains: memory, mood, physical health, living situation and ability to do things for fun. The domains considered most important for quality of life differed between people with dementia, their caregivers and members of the community, with memory the least important domain for all three groups. The rationale for priorities also varied between groups. Conclusions This study confirmed the content validity of the selection of the AD-5D domains and identified multiple differences in the reasons behind stated priorities for quality of life for people with dementia, their caregivers and community members.

Preterm children with suspected cerebral palsy at 19 months corrected age in the Canadian neonatal follow-up network

BackgroundThe ability to definitively diagnose cerebral palsy (CP) at 18–24 months is unknown. AimsTo describe very preterm children who, at 19 months, have suspected CP defined as neither having a definitive diagnosis of CP nor no CP and compare them with children with and without CP. Study design and methodsLongitudinal national cohort study of births <29 weeks' gestation with linked Canadian Neonatal Network and Canadian Neonatal Follow-up Network data with 19 month assessments and 3-year questionnaires (Ages and Stages-3 and Health Status Classification System-Preschool). CP, no CP and suspected CP groups, classified at 19 months, were compared using chi square and ANOVA. ResultsOf 3086 survivors, 2280 had complete 19-month corrected age (CA) and 1261 had 3-year CA data. Suspected CP (3.6%), CP (6.4%) and no CP (90%) groups differed (p < 0.05) in birth weight, gestational age, complications of prematurity and NICU length of stay. Children with suspected CP had Bayley-III motor, cognitive and language composite scores at 18 months midway between CP and no CP, had the lowest sensory impairment rates and highest hospital readmission rates. At 3 years, gross motor, fine motor, problem-solving, communication and social skill abilities differed: abnormal outcomes were intermediate for children with suspected CP (p < 0.01). ConclusionsCP incidence varied from 6.4% to 10% with exclusion or inclusion of children with suspected CP. Children with suspected CP have characteristics mostly midway between those with and without CP and developmental concerns persist to 3 years and require surveillance beyond 19 months.

Patient-Centered Item Selection for a New Preference-Based Generic Health Status Instrument: CS-Base

ObjectivesTo develop patient-centered health content for a novel generic instrument (Château Santé Base [CS-Base]) that is suitable to generate values for health status. MethodsCandidate items were drawn from existing health frameworks of generic health status instruments and placed in a diagram (HealthFANTM, Zeist, the Netherlands). Through an online survey, patients with a wide range of diseases were asked to select the 9 items that were most important to them. The importance of the items for the whole study group was determined by means of frequency distributions. ResultsAfter handling duplicates and overlap, the remaining set of 47 items was placed in the HealthFAN. Among the 2256 Dutch patients who started the survey, the most common diagnoses were neck and back pain, diabetes, and asthma/chronic obstructive pulmonary disease. The 5 health items mentioned most frequently as most important were pain, personal relationships, fatigue, memory, and vision. Hearing and vision, anxiety and depression, and independence and self-esteem seemed highly intertwined, so we chose to pair these items. ConclusionsA total of 12 health items were included in CS-Base. Its content is largely based on patient input and enables classification of patients' health status. CS-Base can be administered by means of an app on a mobile phone, which makes it a convenient and attractive tool for patients and researchers.

Incidence of Prostate Cancer according to Metabolic Health Status: a Nationwide Cohort Study

BackgroundWe assessed the association between metabolic health status and incidence of prostate cancer using the National Health Check-ups (NHC) database of Korea.MethodsA total of 11,771,252 men who participated in the NHC between 2009 and 2012 and 56,552 men who were newly diagnosed with prostate cancer were analyzed. Normal-weight and obesity were defined as body mass index (BMI) < 25 kg/m2 and ≥ 25 kg/m2, respectively. Metabolic obesity was defined as the presence ≥ 3 components of the metabolic syndrome. Participants were stratified into 4 groups: metabolically healthy, normal-weight; metabolically obese, normal-weight (MONW); metabolically healthy, obese (MHO); and metabolically obese, obese. Multivariate Cox regression analysis was performed to examine the relationship between metabolic health status and incidence of prostate cancer.ResultsDuring a mean 5.4 ± 1.1 years of follow-up, 56,552 patients were registered with a diagnosis of prostate cancer. When analyzed according to metabolic health status classification, the multivariable-adjusted hazard ratio (HR) was 1.143 for the MONW group, 1.097 for the MHO group, showing the HR for the MONW group was higher than that for the MHO group. As the number of metabolic syndrome components increased, HR increased significantly. When stratified based on BMI, metabolically obese patients showed significantly higher HR than metabolically healthy patients in all BMI groups.ConclusionThis population-based nationwide study revealed an association between metabolic health status and the incidence of prostate cancer, and the risk increased according to the number of components of the metabolic syndrome.

Feasibility and validity of the Health Status Classification System-Preschool (HSCS-PS) in a large community sample: the Generation R study

ObjectivesTo evaluate the feasibility, discriminant validity and concurrent validity of the Health Status Classification System-Preschool (HSCS-PS) in children aged 3 years in a large community sample in the Netherlands.Design/settingA prospective...

1453. Ninety-One Day Quality of Life Post-Pneumonia Diagnosis in Adult Patients in Japan

BackgroundPneumonia is a serious illness with potentially long-lasting but poorly-characterized impact on quality of life. The Japanese Goto Epidemiology Study is a prospective, active, population-based surveillance study of adults with community-onset pneumonia (COP), that includes assessment of Quality Adjusted Life Years (QALYs).MethodsPatients with X-ray/CT scan confirmed COP enrolled in the Goto study and consented to participate in QALY assessment responded to Japanese versions of EuroQol-5D-5L (EQ-5D-5L) health state classification (primary), EQ-5D visual analog scale, and SF-6D (secondary) instruments. This interim analysis reports 91-day QALYs based on Day 1 (diagnosis), 8, 16, 31, and 91 EQ-5D-5L responses of patients enrolled between June 1, 2017 and February 7, 2008. For comparison, we developed hypothetical QALYs had the patients not developed pneumonia (control) using the EQ-5D-5L scores from Day -30 (via recall) carried forward and adjusted by the natural decline in scores and death with age. QALYs were calculated as the area (trapezoidal method) under the survival weighted pneumonia and control EQ-5D-5L QALY score curves.ResultsThe 234 patients were 55% male, 88% aged ≥64 years, 45% nursing home residents, and 65% initially hospitalized (35% initially outpatient) for COP. Compliance for interviews among survivors was 100%. EQ-5D-5L scores were 0.732 at Day −30, decreased to 0.590 at diagnosis, and rose to 0.675 by Day 91. The average scores at all time points remained below Day −30 (all P values <0.01). Compared with hypothetical controls, development of pneumonia on average resulted in a loss of 0.0292 QALYs (P < 0.001) during the first 91 days of follow-up.ConclusionAmong residents of Goto Island, Japan, significant QALY losses were observed in association with a diagnosis of pneumonia and had not returned to baseline by 3 months after diagnosis. Scores and cumulative QALY losses during the first 3 months after pneumonia diagnosis were comparable to those experienced by US adults with chronic heart failure during a 3-month period.Disclosures H. Glick,K. Hirano, Pfizer Inc.: Consultant, Research support. T. Miyazaki, Pfizer Inc.: Collaborator, Research support and Speaker honorarium. . J. Suaya, E. Gonzalez, B. D. Gessner, A. G. Arguedas, Pfizer Inc.: Employee and Shareholder, Salary. R. E. Isturiz, Pfizer, Inc: Employee and Shareholder, Salary and Stock & Stock Options. . S. Kohno, Pfizer Inc.: Consultant, Research support and Speaker honorarium.

Effectiveness of Tailored Rehabilitation Education in Improving the Health Literacy and Health Status of Postoperative Patients With Breast Cancer: A Randomized Controlled Trial.

The improvement of breast cancer treatment and the extension of survivorship have led to the development of postoperative complications among cancer survivors. Health literacy (HL), defined as patients' capability of using health information to maintain their health status, can enable breast cancer patients to manage postoperative complications. The aims of this study were to develop a tailored rehabilitation education (TRE) program and examine the effectiveness of this program in improving the HL and health status with breast cancer. This randomized controlled trial recruited 99 breast cancer patients (49 and 50 in the intervention and control groups, respectively) within 1 week after surgery. Four-week individualized TRE programs were implemented to improve their HL and health status. Our results showed that the TRE program produced significant improvements in HL and health status in the components of the International Classification of Functioning, Disability and Health. However, no significant difference was observed in the activity scores obtained using the Barthel Index between the 2 groups. Our finding supports the effectiveness of 1-month TRE in improving HL and all components of the International Classification of Functioning, Disability and Health status, except the activity component among breast cancer in Taiwan. Clinicians could incorporate the TRE techniques in the rehabilitation sessions according to the healthcare, disease prevention, and health promotion domains to improve the clinical outcomes as well as change their health behaviors and attitudes of patients with breast cancer.

Insomnia and Impaired Quality of Life in the United States.

This analysis characterizes the individual-level and population-level burden of insomnia in relation to other medical conditions and describes the comorbidity of insomnia with other medical conditions, including the dependence of these comorbidities on pain, life events, and mental disorders. Information from 34,712 adults in the National Epidemiologic Survey on Alcohol and Related Conditions-III (2012-2013) was analyzed. Quality-adjusted life-years (QALYs) were measured with the SF-6D, a 6-dimensional health state classification derived from the Short-Form-12, version 2. In the last 12 months, 27.3% of adults reported insomnia. The US annual loss of QALYs associated with insomnia (5.6 million; 95% CI, 5.33-5.86 million) was significantly larger than that associated with any of the other 18 medical conditions assessed, including arthritis (4.94 million; 95% CI, 4.62-5.26 million), depression (4.02 million; 95% CI, 3.87-4.17 million), and hypertension (3.63 million; 95% CI, 3.32-3.93 million). After control for demographic factors, all conditions examined from obesity (adjusted odds ratio [aOR] = 1.25) to mania (aOR = 5.04) were associated with an increased risk of insomnia. Further controlling for pain, stressful life events, and mental disorders decreased the odds of the co-occurrence of insomnia with these conditions. The decrease in insomnia comorbidity associated with pain was greatest for fibromyalgia (31.8%) and arthritis (20.1%); the decrease in insomnia comorbidity associated with life events was greatest for mania (13.4%) and drug use disorders (11.2%); and the decrease in insomnia comorbidity associated with mental disorders was greatest for peptic ulcer disease (11.2%) and liver diseases (11.1%). Insomnia is prevalent and associated with substantial population-level burden in self-assessed health. The co-occurrence of insomnia with common medical conditions is differentially related to pain and to a lesser extent to stressful life events and mental disorders.

Comparison of two methods for indirect measurement of atmospheric dust deposition: Street-dust composition and vegetation-health status derived from hyperspectral image data.

This study presents a statistical comparison between the in situ measurements of the elemental composition of street dust and a forest health status classification derived from aerial hyperspectral image data (HyMap). Combining these two methods allowed us to indirectly pinpoint at a high spatial resolution the atmospheric dust emissions and its effects in a study area around the open-pit lignite mine in Sokolov, Czech Republic. The results reveal a statistically significant relationship between increased Al, Na, Li and Sr levels in street dust and decreased forest health status, and the highest number of statistically significant correlations within a 100m distance from the street-dust sampling points. Differences in lithological composition were unable to sufficiently explain these changes, therefore anthropogenic factors like dust emissions from coal mining and coal combustion, as well as urbanisation and other industries might be the reason for this link. Such studies are a crucial step in developing new high spatial resolution methods for determining atmospheric dust deposition and their effects.

Modelling a preference-based index for EQ-5D using a non-parametric Bayesian method.

Conventionally, models used for health state valuation data have been parametric. Recently, a number of researchers have investigated the use of non-parametric Bayesian methods in this area. In this paper, we present a non-parametric Bayesian model to estimate a preference-based index for a five-dimensional health state classification, namely EQ-5D. A sample of 2997 members of the UK general population valued 43 health states selected from a total of 243 health states defined by the EQ-5D using time trade-off technique. Findings from non-parametric modelling are reported in this paper and compared to previously used parametric estimations. The impact of respondent characteristics on health state valuations is also reported. The non-parametric models were found to be better at predicting scores in populations with different distributions of characteristics than observed in the survey sample. Additionally, non-parametric models were found to be better at allowing for the impact of respondent characteristics to vary by health state. The results show an important age effect with sex having some effect. The non-parametric Bayesian models provide more realistic and better utility estimates from the EQ-5D than previously used parametric models have done. Furthermore, the model is more flexible in estimating the impact of covariates.

A preference-based item response theory model to measure health: concept and mathematics of the multi-attribute preference response model

BackgroundA new patient-reported health measurement model has been developed to quantify descriptions of health states. Known as the multi-attribute preference response (MAPR) model, it is based on item response theory. The response task in the MAPR is for a patient to judge whether hypothetical health-state descriptions are better or worse than his/her own health status.MethodsIn its most simple form MAPR is a Rasch model where for each respondent on the same unidimensional health scale values are estimated of their own health status and values of the hypothetical comparator health states. These values reflect the quality or severity of the health states. Alternatively, the respondents are offered health-state descriptions that are based on a classification system (e.g., multi-attribute) with a fixed number of health attributes, each with a limited number of levels. In the latter variant, the weights of the levels of the attributes in the descriptive system, which represents the range of the health states, are estimated. The results of a small empirical study are presented to illustrate the procedures of the MAPR model and possible extensions of the model are discussed.ResultsThe small study that we conducted to illustrate the procedure and results of our proposed method to measure the quality of health states and patients’ own health status showed confirming results.ConclusionsThis paper introduces the typical MAPR model and shows how it extends the basic Rasch model with a regression function for the attributes of the health-state classification system.

Developing a dementia-specific preference-­based quality of life measure (AD-5D) in Australia: a valuation study protocol

IntroductionGeneric instruments for assessing health-related quality of life may lack the sensitivity to detect changes in health specific to certain conditions, such as dementia. The Quality of Life in Alzheimer’s...

Aquatic ecosystem health and trophic status classification of the Bitter Lakes along the main connecting link between the Red Sea and the Mediterranean.

The Bitter Lakes are the most significant water bodies of the Suez Canal, comprising 85% of the water volume, but spreading over only 24% of the length of the canal. The present study aims at investigation of the trophic status of the Bitter Lakes employing various trophic state indices, biotic and abiotic parameters, thus reporting the health of the Lake ecosystem according to the internationally accepted classification criteria’s. The composition and abundance of phytoplankton with a dominance of diatoms and a decreased population density of 4315–7376 ind. l−1 reflect the oligotrophic nature of this water body. The intense growth of diatoms in the Bitter Lakes depends on silicate availability, in addition to nitrate and phosphate. If the trophic state index (TSI) is applied to the lakes under study it records that the Bitter Lakes have an index under 40. Moreover, in the total chlorophyll-a measurements of 0.35–0.96 µg l−1 there are more indicative of little algal biomass and lower biological productivity. At 0.76–2.3 µg l−1, meanwhile, the low quantity of Phosphorus is a further measure of low biological productivity. In the Bitter Lakes, TN/TP ratios are high and recorded 147.4, and 184.7 for minimum and maximum ratios, respectively. These values indicate that in Bitter lakes, the limiting nutrient is phosphorus and confirm the oligotrophic status of the Bitter Lakes. The latter conclusion is supported by Secchi disc water clarity measurements, showing that light can penetrate, and thus algae can photosynthesize, as deep as >13 m. This study, therefore, showed that the Bitter Lakes of the Suez Canal exhibit oligotrophic conditions with clear water, low productivity and with no algal blooming.

Utility and quality-adjusted life-years in coronary artery disease: Five-year follow-up of the MASS II trial.

This study evaluated the utility and quality-of-life year measurements for patients with coronary artery disease who underwent any of 3 therapeutic strategies with a 5-year follow-up. Quality-of-life data were obtained from the Medicine, Angioplasty, or Surgery Study II trial. To obtain utilities, the 36-Item Short-Form questionnaire was converted to a 6-Dimensional Health State Classification System. Of the 611 initial patients, 579 completed the questionnaire. In all, 188 patients received the surgical treatment-194 the percutaneous, and the remaining 197 the medical. The median utility scores for the 5 years analyzed were 0.809 (95% confidence interval [CI] 0.794-0.842) for patients assigned to percutaneous coronary intervention, 0.755 (95% CI 0.723-0.774) for medical treatment, and 0.780 (95% CI 0.761-0.809) for coronary artery bypass graft surgery. The difference between percutaneous coronary intervention and medical treatment was statistically significant (P < .05, Dunn test). The median cumulative quality-of-life years across the 5 years were 3.802 (95% CI 3.668-3.936) for percutaneous, 3.540 (95% CI 3.399-3.681) for medical, and 3.764 (95% CI 3.638-3.890) for surgery. Additionally, the median quality-of-life years between percutaneous and medical treatment was 0.262 (95% CI 0.068-0.456), between surgery and medical treatment it was 0.224 (95% CI 0.036-0.413), and between surgery and percutaneous coronary intervention it was -0.038 (95% CI -0.221 to -0.146). Coronary artery bypass surgery and percutaneous coronary intervention were similar regarding cumulative quality-of-life years; however, they were both superior to that of medical treatment. The results presented are valuable data for further cost-utility studies.

A Review of Generic Preference-Based Measures for Use in Cost-Effectiveness Models

Generic preference-based measures (GPBMs) of health are used to obtain the quality adjustment weight required to calculate the quality-adjusted life year in health economic models. GPBMs have been developed to use across different interventions and medical conditions and typically consist of a self-complete patient questionnaire, a health state classification system, and preference weights for all states defined by the classification system. Of the six main GPBMs, the three most frequently used are the Health Utilities Index version 3, the EuroQol 5 dimensions (3 and 5 levels), and the Short Form 6 dimensions. There are considerable differences in GPBMs in terms of the content and size of descriptive systems (i.e. the numbers of dimensions of health and levels of severity within these), the methods of valuation [e.g. time trade-off (TTO), standard gamble (SG)], and the populations (e.g. general population, patients) used to value the health states within the descriptive systems. Although GPBMs are anchored at 1 (full health) and 0 (dead), they produce different health state utility values when completed by the same patient. Considerations when selecting a measure for use in a clinical trial include practicality, reliability, validity and responsiveness. Requirements of reimbursement agencies may impose additional restrictions on suitable measures for use in economic evaluations, such as the valuation technique (TTO, SG) or the source of values (general public vs. patients).

Isolated neonatal MRI punctate white matter lesions in very preterm neonates and quality of life at school age.

To study the quality of life at school age of very preterm infants presenting isolated punctate periventricular white matter lesions (IPWL) on late-preterm or term magnetic resonance imaging (MRI). In 1996-2000, 16 of the 131 very preterm neonates explored by MRI were found to have IPWL. At the age of 9-14, 12 children from the IPWL group were compared with 54 children born preterm but with a normal MRI (no lesion). Quality of life (Health Status Classification System Pre School questionnaire), school performance, and motor outcome were investigated. Overall quality of life did not differ between the groups (classified as perfect in 2/12 of the IPWL vs 20/54 in the no-lesion). The sub-items mobility and dexterity differed significantly between the two groups, with impairment in the IPWL group (p < 0.001 and p < 0.05). This group also displayed higher levels of motor impairment: they began walking later [20(4) vs. 15(3) months), p < 0.01], had higher frequencies of cerebral palsy (6/12 vs. 2/54, p < 0.05), and dyspraxia (4/12 vs. 0/54, p < 0.001). The rate of grade retention did not differ between the groups (3/12 in the IPWL group vs. 17/54 in the no-lesions group) but, as expected, was higher than that of the French general population (17.4%) during the study period. This long-term follow-up study detected no increase in the risk of subsequent cognitive impairment in very preterm infants with IPWL, but suggests that these children may have a significantly higher risk of dyspraxia, and motor impairment.

Experience-based utility and own health state valuation for a health state classification system: why and how to do it

In the estimation of population value sets for health state classification systems such as the EuroQOL five dimensions questionnaire (EQ-5D), there is increasing interest in asking respondents to value their own health state, sometimes referred to as “experience-based utility values” or, more correctly, own rather than hypothetical health states. Own health state values differ to hypothetical health state values, and this may be attributable to many reasons. This paper critically examines whose values matter; why there is a difference between own and hypothetical values; how to measure own health state values; and why to use own health state values. Finally, the paper examines other ways that own health state values can be taken into account, such as including the use of informed general population preferences that may better take into account experience-based values.

Developing preference-based measures for diabetes: DHP-3D and DHP-5D.

The aim of this study was to develop two diabetes-specific preference-based measures [the Diabetes Health Profile-3 Dimension (DHP-3D) and the Diabetes Health Profile-5 Dimension (DHP-5D)] for use in the calculation of Quality Adjusted Life Years, a key outcome in economic evaluation. These measures were based on the non-preference-based instrument the Diabetes Health Profile. For DHP-3D, psychometric and Rasch analyses were used to develop a health state classification system based on the Diabetes Health Profile-18 (DHP-18). The DHP-5D added two dimensions to the DHP-3D to extend the range of impacts measured. Each classification system was valued by 150 general public respondents in the United Kingdom using Time Trade Off (TTO). Multivariate regression was used to estimate utility value sets. The matched dimensions across each measure were compared using z-score tests. The DHP-3D included three dimensions defined as mood, eating and social limitations, and the DHP-5D added dimensions defined as hypoglycaemic attacks and vitality. For both, the random effects generalized least squares regression model produced consistent value sets, with the DHP-3D and DHP-5D ranging from 0.983 (best state) to 0.717 (worst state), and 0.979 to 0.618 respectively. The addition of the two extra dimensions leads to significant differences for the more severe levels of each matched dimension. We have developed two diabetes-specific preference-based measures that, subject to psychometric assessment, can be used to provide condition-specific utility values to complement generic utilities from more widely validated measures such as the EuroQol-5 Dimension.

Development of a Patient-Reported Palliative Care-Specific Health Classification System: The POS-E.

BackgroundGeneric preference-based measures are commonly used to estimate quality-adjusted life-years (QALYs) to inform resource-allocation decisions. However, concerns have been raised that generic measures may be inappropriate in palliative care.ObjectiveOur objective was to derive a health-state classification system that is amenable to valuation from the ten-item Palliative Care Outcome Scale (POS), a widely used patient-reported outcome measure in palliative care.MethodsThe dimensional structure of the original POS was assessed using factor analysis. Item performance was assessed, using Rasch analysis and psychometric criteria, to enable the selection of items that represent the dimensions covered by the POS. Data from six studies of patients receiving palliative care were combined (N = 1011) and randomly split into two halves for development and validation. Analysis was undertaken on the development data, and results were validated by repeating the analysis with the validation dataset.ResultsFollowing Rasch and factor analyses, a classification system of seven items was derived. Each item had two to three levels. Rasch threshold map helped identify a set of 14 plausible health states that can be used for the valuation of the instrument to derive a preference-based index.ConclusionCombining factor analysis and Rasch analysis with psychometric criteria provides a valid method of constructing a classification system for a palliative care-specific preference-based measure. The next stage is to obtain preference weights so the measure can be used in economic evaluations in palliative care.Electronic supplementary materialThe online version of this article (doi:10.1007/s40271-017-0224-1) contains supplementary material, which is available to authorized users.