Somtimes: self organizing maps for time series clustering and its application to serious illness conversations.

There is demand for scalable algorithms capable of clustering and analyzing large time series data. The Kohonen self-organizing map (SOM) is an unsupervised artificial neural network for clustering, visualizing, and reducing the dimensionality of complex data. Like all clustering methods, it requires a measure of similarity between input data (in this work time series). Dynamic time warping (DTW) is one such measure, and a top performer that accommodates distortions when aligning time series. Despite its popularity in clustering, DTW is limited in practice because the runtime complexity is quadratic with the length of the time series. To address this, we present a new a self-organizing map for clustering TIME Series, called SOMTimeS, which uses DTW as the distance measure. The method has similar accuracy compared with other DTW-based clustering algorithms, yet scales better and runs faster. The computational performance stems from the pruning of unnecessary DTW computations during the SOM's training phase. For comparison, we implement a similar pruning strategy for K-means, and call the latter K-TimeS. SOMTimeS and K-TimeS pruned 43% and 50% of the total DTW computations, respectively. Pruning effectiveness, accuracy, execution time and scalability are evaluated using 112 benchmark time series datasets from the UC Riverside classification archive, and show that for similar accuracy, a 1.8× speed-up on average for SOMTimeS and K-TimeS, respectively with that rates vary between 1× and 18× depending on the dataset. We also apply SOMTimeS to a healthcare study of patient-clinician serious illness conversations to demonstrate the algorithm's utility with complex, temporally sequenced natural language. Supplementary Information: The online version contains supplementary material available at 10.1007/s10618-023-00979-9.

Patient-Caregiver Dyads & End-of-Life Care: Caregiver Personality Disrupts Gender-Based Norms.

CONTEXT: Gender and personality may individually impact end-of-life care. Men often receive more aggressive treatments than women near death, and personality - particularly openness - may be associated with increased care utilization when it diverges from traditional treatment norms. However, research has not examined the interaction of these variables in a dyadic context. OBJECTIVES: This study examined the dyadic effects of patient gender and caregiver personality on end-of-life care. METHODS: Using data from the VOICE randomized clinical trial, the present sample consisted of patient-caregiver dyads receiving outpatient care for advanced cancer in Sacramento and Western New York. Analyses assessed whether caregiver personality was associated with gender-based differences in patient chemotherapy or emergency department/inpatient visits in the 30 days before death. Logistic regression examined the interaction between caregivers' Big Five personality dimensions and patient gender while accounting for patient and caregiver demographic and health characteristics. RESULTS: Of a total of 134 patient-caregiver dyads, 19.4% (n = 26) of patients received chemotherapy and 47.8% (n = 64) had an emergency department/inpatient visit in the 30 days before death. Results demonstrated a significant interaction between caregiver openness and patient gender on receipt of chemotherapy (odds ratio = 0.07, p = 0.006). When caregivers were less open, men were more likely to receive chemotherapy near death, whereas when caregivers were more open, women were more likely to receive chemotherapy near death. CONCLUSION: Results suggest caregiver personality characteristics, particularly openness, might disrupt gender-based treatment norms at end-of-life. Findings demonstrate that patient and caregiver factors can interact to explain patient healthcare utilization.

Protocol for a Scalable StoryListening Intervention for Grief-Related Loneliness During COVID-19.

Background: Social distancing during the COVID-19 pandemic limited how family, friends, and clinicians physically interacted with people who were dying and decreased communal opportunities for processing grief. These barriers can cause or exacerbate suffering due to loneliness while grieving. Purpose: In this article, we describe the protocol for a brief storytelling intervention designed to reduce loneliness among families, friends, and clinicians grieving the death of a person during the time of COVID-19. Methods: We trained four StoryListening doulas (SLDs) to hold a welcoming space and listen to stories with curiosity and openness. The intervention included a video StoryListening session and two brief questionnaires, filled out before and two weeks after the encounter, assessing loneliness and quality of life. During sessions, SLDs invited participants to share their story of loss in their own words and in as much detail as preferred. When participants felt a sense of story completion, SLDs shared validating statements and expressed gratitude to the participant for sharing. The video and audio for each participant's StoryListening encounter were recorded and the participant was offered an audio copy of their session.

Developing a Direct Observation Measure of Therapeutic Connection in Psilocybin-Assisted Therapy: A Feasibility Study.

Context: Measuring therapeutic connection during psilocybin-assisted therapy is essential to understand underlying mechanisms, inform training, and guide quality improvement. Purpose: To evaluate the feasibility of directly observing indicators of therapeutic connection during psilocybin administration encounters. Methods: We evaluated audio and video data from a recent clinical trial for observable expressions of therapeutic connection as defined in proposed best-practice competencies (i.e., empathic abiding presence and interpersonal grounding). We selected the first four 8-hour encounters involving unique participants, therapists, and gender pairs. Each video was independently coded by three members of an interprofessional six-person team. Using a structured checklist, coders recorded start-stop times, the audible (i.e., speech prosody or words) and visible (i.e., body movements, eye gaze, and touch) cues marking the event, and the qualities of the interaction (e.g., expression of awe, trust, distress, and calmness). We assessed feasibility by observing the frequency, distribution, and overlap of cues and qualities coders used to identify and define moments of therapeutic connection. Results: Among the 2074 minutes of video, coders recorded 372 moments of therapeutic connection. Eighty-three percent were identified by at least two coders and 41% by all three. Coders used a combination of audible and visual cues to identify therapeutic connection in 51% of observed events (190/372). Both the cues and qualities of therapeutic connection expressions varied over the course of psilocybin temporal effects on states of consciousness. Conclusion: Direct observation of therapeutic human connection is feasible, sensitive to changes in states of consciousness and requires evaluation of audible and visual data.

An Acoustical and Lexical Machine-Learning Pipeline to Identify Connectional Silences.

Context: Developing scalable methods for conversation analytics is essential for health care communication science and quality improvement. Purpose: To assess the feasibility of automating the identification of a conversational feature, Connectional Silence, which is associated with important patient outcomes. Methods: Using audio recordings from the Palliative Care Communication Research Initiative cohort study, we develop and test an automated measurement pipeline comprising three machine-learning (ML) tools-a random forest algorithm and a custom convolutional neural network that operate in parallel on audio recordings, and subsequently a natural language processing algorithm that uses brief excerpts of automated speech-to-text transcripts. Results: Our ML pipeline identified Connectional Silence with an overall sensitivity of 84% and specificity of 92%. For Emotional and Invitational subtypes, we observed sensitivities of 68% and 67%, and specificities of 95% and 97%, respectively. Conclusion: These findings support the capacity for coordinated and complementary ML methods to fully automate the identification of Connectional Silence in natural hospital-based clinical conversations.

Methodological innovations to strengthen evidence-based serious illness communication.

BACKGROUND/OBJECTIVE: A growing population of those affected by serious illness, prognostic uncertainty, patient diversity, and healthcare digitalization pose challenges for the future of serious illness communication. Yet, there is paucity of evidence to support serious illness communication behaviors among clinicians. Herein, we propose three methodological innovations to advance the basic science of serious illness communication. RESULTS: First, advanced computation techniques - e.g. machine-learning techniques and natural language processing - offer the possibility to measure the characteristics and complex patterns of audible serious illness communication in large datasets. Second, immersive technologies - e.g., virtual- and augmented reality - allow for experimentally manipulating and testing the effects of specific communication strategies, and interactional and environmental aspects of serious illness communication. Third, digital-health technologies - e.g., shared notes and videoconferences - can be used to unobtrusively observe and manipulate communication, and compare in-person to digitally-mediated communication elements and effects. Immersive and digital health technologies allow integration of physiological measurement (e.g. synchrony or gaze) that may advance our understanding of patient experience. CONCLUSION/PRACTICE IMPLICATIONS: New technologies and measurement approaches, while imperfect, will help advance our understanding of the epidemiology and quality of serious illness communication in an evolving healthcare environment.

Integration of palliative care consultation into the management of patients with chronic limb-threatening ischemia.

OBJECTIVE: We assessed the feasibility of integrating palliative care consultation into the routine management of patients with chronic limb-threatening ischemia (CLTI). Additionally, we sought to describe patient-reported outcomes from the palliative care and vascular literature in patients with CLTI receiving a palliative care consultation at our institution. METHODS: This was a single-institution, prospective, observational study that aimed to assess feasibility of incorporating palliative care consultation into the management of patients admitted to our tertiary academic medical center with CLTI by looking at utilization of palliative care before and after implementation of a protocol-based palliative care referral system. A survey comprised of patient-reported outcomes from the palliative care literature was administered to patients before and after palliative consultation. Length of stay and mortality were compared between our study cohort and a historic cohort of patients admitted with CLTI. RESULTS: Over a 14-month enrollment period, 44% of patients (n = 39) with CLTI (rest pain, 36%; tissue loss, 64%) admitted to the vascular service received palliative care consultation, compared with 5% of patients (n = 4) who would have met criteria over the preceding 14 months before our protocol was instituted. The mean age was 69 years, 23% were female, 92% were white, and 49% were able to ambulate independently. Revascularization included bypass (46%), peripheral vascular intervention (23%), and femoral endarterectomy (21%). Additional procedures included minor amputation or wound debridement (26%) and major amputation (15%). No patients received medical management alone. After receiving palliative care consultation, patients reported experiencing less emotional distress than before consultation (P = .03). They also reported being less bothered by uncertainty regarding what to expect from the course of their illness (P = .002). Fewer patients reported being unsure of the purpose of their medical care after palliative care consultation (8%) vs before (18%), although this was not statistically significant (P = .10). Median length of stay was longer in the study group compared with the historic cohort (8 vs 7 days; P = .02). There was no difference in 30-day mortality (3% vs 8%; P = .42) between the study group and the historic cohort (n = 77). CONCLUSIONS: Integrating inpatient palliative care consultation into the routine management of patients with CLTI is feasible and may improve emotional domains of health-related quality of life. This study laid the foundation for future studies on longer term outcomes of patients with CLTI undergoing palliative care consultation as well as the benefit of outpatient palliative care consultation in patients with CLTI.

Sentiment analysis of medical record notes for lung cancer patients at the Department of Veterans Affairs.

Natural language processing of medical records offers tremendous potential to improve the patient experience. Sentiment analysis of clinical notes has been performed with mixed results, often highlighting the issue that dictionary ratings are not domain specific. Here, for the first time, we re-calibrate the labMT sentiment dictionary on 3.5M clinical notes describing 10,000 patients diagnosed with lung cancer at the Department of Veterans Affairs. The sentiment score of notes was calculated for two years after date of diagnosis and evaluated against a lab test (platelet count) and a combination of data points (treatments). We found that the oncology specific labMT dictionary, after re-calibration for the clinical oncology domain, produces a promising signal in notes that can be detected based on a comparative analysis to the aforementioned parameters.

The StoryListening Project: Feasibility and Acceptability of a Remotely Delivered Intervention to Alleviate Grief during the COVID-19 Pandemic.

Background: The events surrounding the COVID-19 pandemic have created heightened challenges to coping with loss and grief for family and friends of deceased individuals, as well as clinicians who experience loss of their patients. There is an urgent need for remotely delivered interventions to support those experiencing grief, particularly due to growing numbers of bereaved individuals during the COVID-19 pandemic. Objective: To determine the feasibility and acceptability of the brief, remotely delivered StoryListening storytelling intervention for individuals experiencing grief during the COVID pandemic. Setting/Subjects: A single-arm pilot study was conducted in the United States. Participants included adult English-speaking family members, friends, or clinicians of individuals who died during the COVID-19 pandemic. All participants engaged in a televideo StoryListening session with a trained StoryListening doula. Measurements: Participants completed a brief follow-up telephone interview two weeks after the StoryListening session. We describe enrollment and retention data to assess feasibility and conducted a deductive thematic analysis of the follow-up interview data to assess acceptability. Results: Sixteen clinicians and 48 friends/family members enrolled in the study (n = 64; 75% enrollment), 62 completed a StoryListening session; 60 completed the follow-up interview. Participants reported that the intervention was useful and offered a valuable opportunity to process their grief experience. Conclusions: The StoryListening intervention is feasible and acceptable for friends/family members and clinicians who have experienced grief during COVID. Our intervention may offer an accessible first-line option to address the increasing wave of bereavement-related distress and clinician burnout in the United States.

A Formative Mixed-Methods Study of Emotional Responsiveness in Telepalliative Care.

Background: It is unknown whether telemedicine-delivered palliative care (tele-PC) supports emotionally responsive patient-clinician interactions. Objectives: We conducted a mixed-methods formative study at two academic medical centers in rural U.S. states to explore the acceptability, feasibility, and emotional responsiveness of tele-PC. Design: We assessed clinicians' emotional responsiveness through questionnaires, qualitative interviews, and video coding. Results: We completed 11 tele-PC consultations. Mean age was 71 years, 30% did not complete high school, 55% experienced at least moderate financial insecurity, and 2/3 rated their overall health poorly. All patients rated tele-PC as equal to, or better than, in-person PC at providing emotional support. There was a tendency toward higher positive and lower negative emotions following the consultation. Video coding identified 114 instances of patients expressing emotions, and clinicians detected and responded to 98% of these events. Conclusion: Tele-PC appears to support emotionally responsive patient-clinician interactions. A mixed-methods approach to evaluating tele-PC yields useful, complementary insights.

Toward a basic science of communication in serious illness.

High-quality communication can mitigate suffering during serious illness. Innovations in theory and technology present the opportunity to advance serious illness communication research, moving beyond inquiry that links broad communication constructs to health outcomes toward operationalizing and understanding the impact of discrete communication functions on human experience. Given the high stakes of communication during serious illness, we see a critical need to develop a basic science approach to serious illness communication research. Such an approach seeks to link "what actually happens during a conversation" - the lexical and non-lexical communication content elements, as well as contextual factors - with the emotional and cognitive experiences of patients, caregivers, and clinicians. This paper defines and justifies a basic science approach to serious illness communication research and outlines investigative and methodological opportunities in this area. A systematic understanding of the building blocks of serious illness communication can help identify evidence-informed communication strategies that promote positive patient outcomes, shape more targeted communication skills training for clinicians, and lead to more tailored and meaningful serious illness care.

Content of Tele-Palliative Care Consultations with Patients Receiving Dialysis.

Background: Little is known about the content of communication in palliative care telehealth conversations in the dialysis population. Understanding the content and process of these conversations may lead to insights about how palliative care improves quality of life. Methods: We conducted a qualitative analysis of video recordings obtained during a pilot palliative teleconsultation program. We recruited patients receiving dialysis from five facilities affiliated with an academic medical center. Palliative care clinicians conducted teleconsultation using a wall-mounted screen with a camera mounted on a pole and positioned mid-screen in the line of sight to facilitate direct eye contact. Patients used an iPad that was attached to an IV pole positioned next to the dialysis chair. Conversations were coded using a preexisting framework of themes and content from the Serious Illness Conversation Guide (SICG) and revised Edmonton Symptom Assessment System-Renal. Results: We recruited 39 patients to undergo a telepalliative care consultation while receiving dialysis, 34 of whom completed the teleconsultation. Specialty palliative care clinicians (3 physicians and 1 nurse practitioner) conducted 35 visits with 34 patients. Median (interquartile range) duration of conversation was 42 (28-57) minutes. Most frequently discussed content included sources of strength (91%), critical abilities (88%), illness understanding (85%), fears and worries (85%), what family knows (85%), fatigue (77%), and pain (65%). Process features such as summarizing statements (85%) and making a recommendation (82%) were common, whereas connectional silence (56%), and emotion expression (21%) occurred less often. Conclusions: Unscripted palliative care conversations in outpatient dialysis units through telemedicine exhibited many domains recommended by the SICG, with less frequent discussion of symptoms. Emotion expression was uncommon for these conversations that occurred in an open setting.

A novel Scale to Assess Palliative Care Patients' Experience of Feeling Heard and Understood.

CONTEXT: Patient experience of palliative care serves as an important indicator of quality and patient-centeredness. OBJECTIVES: To develop a novel patient-reported scale measuring ambulatory palliative care patients' experience of feeling heard and understood by their providers. METHODS: We used self-reported patient experience data collected via mixed-mode survey administration. We conducted an exploratory factor analysis (EFA) and an expert panel ranking exercise to reduce the 10-item set based on underlying dimensionality. We then used item response theory (IRT) to calibrate remaining items based on psychometric properties and test information and precision. We considered item-level fit and examined the standardized local dependence chi-square statistics. We evaluated candidate items for differential item functioning by survey mode. We evaluated the test-retest reliability and validity of the final scale. RESULTS: The EFA yielded a single factor (9/10 items had loadings > 0.80 on the single factor). We removed two items with the lowest factor loadings and ranked by the expert panel as being least reflective of the overall construct. IRT calibration of the remaining eight items showed high slopes (range 2.66 - 5.18); location parameters were all negative (range -0.90 - -0.36). We removed two more items based on local dependence indices and item-level fit. Combining psychometric information with the expert ratings we established the final 4-item scale, which was reliable (Cronbach's alpha = 0.84; polychoric correlation coefficient = 0.72) and had good convergent validity. CONCLUSIONS: This novel multi-item Feeling Heard and Understood scale can be used to measure and improve ambulatory palliative care patient experience.

Patient, caregiver, and oncologist predictions of quality of life in advanced cancer: Accuracy and associations with end-of-life care and caregiver bereavement.

BACKGROUND: Informed treatment decision-making necessitates accurate prognostication, including predictions about quality of life. AIMS: We examined whether oncologists, patients with advanced cancer, and caregivers accurately predict patients' future quality of life and whether these predictions are prospectively associated with end-of-life care and bereavement. MATERIALS & METHODS: We conducted secondary analyses of clinical trial data. Patients with advanced cancer (n = 156), caregivers (n = 156), and oncologists (n = 38) predicted patient quality of life 3 months into the future. Patients subsequently rated their quality of life 3 months later. Medical record data documented chemotherapy and emergency department (ED)/inpatient visits in the 30 days before death (n = 79 decedents). Caregivers self-reported on depression, anxiety, grief, purpose, and regret 7-months post-mortem. In mixed-effects models, patient, caregiver, and oncologist quality-of-life predictions at study entry were used to predict end-of-life care and caregiver outcomes, controlling for patients' quality of life at 3-month follow-up, demographic and clinical characteristics, and nesting within oncologists. RESULTS: Caregivers (P < 0.0001) and oncologists (P = 0.001) predicted lower quality of life than what patients actually experienced. Among decedents, 24.0% received chemotherapy and 54.5% had an ED/inpatient visit. When caregivers' predictions were more negative, patients were less likely to receive chemotherapy (P = 0.028) or have an ED/inpatient visit (P = 0.033), and caregivers reported worse depression (P = 0.002), anxiety (P = 0.019), and grief (P = 0.028) and less purpose in life (P < 0.001) 7-months post-mortem. CONCLUSION: When caregivers have more negative expectations about patients' quality of life, patients receive less intensive end-of-life care, and caregivers report worse bereavement outcomes.

Epidemiology of Connectional Silence in specialist serious illness conversations.

CONTEXT: Human connection can reduce suffering and facilitate meaningful decision-making amid the often terrifying experience of hospitalization for advanced cancer. Some conversational pauses indicate human connection, but we know little about their prevalence, distribution or association with outcomes. PURPOSE: To describe the epidemiology of Connectional Silence during serious illness conversations in advanced cancer. METHODS: We audio-recorded 226 inpatient palliative care consultations at two academic centers. We identified pauses lasting 2+ seconds and distinguished Connectional Silences from other pauses, sub-categorized as either Invitational (ICS) or Emotional (ECS). We identified treatment decisional status pre-consultation from medical records and post-consultation via clinicians. Patients self-reported quality-of-life before and one day after consultation. RESULTS: Among all 6769 two-second silences, we observed 328 (4.8%) ECS and 240 (3.5%) ICS. ECS prevalence was associated with decisions favoring fewer disease-focused treatments (ORadj: 2.12; 95% CI: 1.12, 4.06). Earlier conversational ECS was associated with improved quality-of-life (p = 0.01). ICS prevalence was associated with clinicians' prognosis expectations. CONCLUSIONS: Connectional Silences during specialist serious illness conversations are associated with decision-making and improved patient quality-of-life. Further work is necessary to evaluate potential causal relationships. PRACTICE IMPLICATIONS: Pauses offer important opportunities to advance the science of human connection in serious illness decision-making.

Conversational stories & self organizing maps: Innovations for the scalable study of uncertainty in healthcare communication.

BACKGROUND: Understanding uncertainty in participatory decision-making requires scientific attention to interaction between what actually happens when patients, families and clinicians engage one another in conversation and the multi-level contexts in which these occur. Achieving this understanding will require conceptually grounded and scalable methods for use in large samples of people representing diversity in cultures, speaking and decision-making norms, and clinical situations. DISCUSSION: Here, we focus on serious illness and describe Conversational Stories as a scalable and conceptually grounded framework for characterizing uncertainty expression in these clinical contexts. Using actual conversations from a large direct-observation cohort study, we demonstrate how natural language processing and unsupervised machine learning methods can reveal underlying types of uncertainty stories in serious illness conversations. CONCLUSIONS: Conversational Storytelling offers a meaningful analytic framework for scalable computational methods to study uncertainty in healthcare conversations.

A general model of conversational dynamics and an example application in serious illness communication.

Conversation has been a primary means for the exchange of information since ancient times. Understanding patterns of information flow in conversations is a critical step in assessing and improving communication quality. In this paper, we describe COnversational DYnamics Model (CODYM) analysis, a novel approach for studying patterns of information flow in conversations. CODYMs are Markov Models that capture sequential dependencies in the lengths of speaker turns. The proposed method is automated and scalable, and preserves the privacy of the conversational participants. The primary function of CODYM analysis is to quantify and visualize patterns of information flow, concisely summarized over sequential turns from one or more conversations. Our approach is general and complements existing methods, providing a new tool for use in the analysis of any type of conversation. As an important first application, we demonstrate the model on transcribed conversations between palliative care clinicians and seriously ill patients. These conversations are dynamic and complex, taking place amidst heavy emotions, and include difficult topics such as end-of-life preferences and patient values. We use CODYMs to identify normative patterns of information flow in serious illness conversations, show how these normative patterns change over the course of the conversations, and show how they differ in conversations where the patient does or doesn't audibly express anger or fear. Potential applications of CODYMs range from assessment and training of effective healthcare communication to comparing conversational dynamics across languages, cultures, and contexts with the prospect of identifying universal similarities and unique "fingerprints" of information flow.

"If it's the time, it's the time": Existential communication in naturally-occurring palliative care conversations with individuals with advanced cancer, their families, and clinicians.

OBJECTIVE: To explore how patients with advanced cancer, their families, and palliative care clinicians communicate about existential experience during palliative care conversations. METHODS: We analyzed data from the Palliative Care Communication Research Initiative (PCCRI) - a multisite cohort study conducted between 2014 and 2016 involving hospitalized adults with advanced cancer who were referred for inpatient palliative care consultations at two academic medical centers. We used a qualitative descriptive approach paired with inductive content analysis to analyze a random subsample of 30 patients from the PCCRI study (contributing to 38 palliative care conversations). RESULTS: We found existential communication to be woven throughout palliative care conversations, with key themes related to: 1) time as a pressing boundary; 2) maintaining a coherent self; and 3) connecting with others. CONCLUSION: Communication about existential experience is omnipresent and varied in palliative care conversations between individuals with advanced cancer, their families, and clinicians. PRACTICE IMPLICATIONS: Clinicians can recognize that discussion of time, routines of daily life, and relationships in the clinical context may hold profound existential relevance in palliative care conversations. Understanding how patients and families talk about existential experience in conversation can create opportunities for clinicians to better meet these needs.

"I Had a Lot More Planned": The Existential Dimensions of Prognosis Communication with Adults with Advanced Cancer.

Background: Communication about prognosis is a key ingredient of effective palliative care. When patients with advanced cancer develop increased prognostic understanding, there is potential for existential distress to occur. However, the existential dimensions of prognosis communication are underexplored. Objective: To describe the existential dimensions of prognosis communication in naturally-occurring palliative care conversations. Methods: This study was an explanatory sequential mixed methods design. We analyzed a random subset of patients from the Palliative Care Communication Research Initiative (PCCRI) parent study (n = 34, contributing to 45 palliative care conversations). Data were based on audio-recorded and transcribed inpatient palliative care conversations between adults with advanced cancer, their families, and palliative care clinicians. We stratified the study sample by levels of prognosis communication, and qualitatively examined patterns of existential communication, comparing the intensity, frequency, and content, within and across levels. Results: Existential communication was more common, and of stronger intensity, within conversations with higher levels of prognosis communication. Conversations with more prognosis communication appeared to exhibit a shift toward the existential and away from the more physical nature of the serious illness experience. Conclusion: Existential and prognosis communication are intimately linked within palliative care conversations. Results highlight the multiplicity and mutuality of concerns that arise when contemplating mortality, drawing attention to areas of palliative care communication that warrant future research.

Using natural language processing to explore heterogeneity in moral terminology in palliative care consultations.

BACKGROUND: High quality serious illness communication requires good understanding of patients' values and beliefs for their treatment at end of life. Natural Language Processing (NLP) offers a reliable and scalable method for measuring and analyzing value- and belief-related features of conversations in the natural clinical setting. We use a validated NLP corpus and a series of statistical analyses to capture and explain conversation features that characterize the complex domain of moral values and beliefs. The objective of this study was to examine the frequency, distribution and clustering of morality lexicon expressed by patients during palliative care consultation using the Moral Foundations NLP Dictionary. METHODS: We used text data from 231 audio-recorded and transcribed inpatient PC consultations and data from baseline and follow-up patient questionnaires at two large academic medical centers in the United States. With these data, we identified different moral expressions in patients using text mining techniques. We used latent class analysis to explore if there were qualitatively different underlying patterns in the PC patient population. We used Poisson regressions to analyze if individual patient characteristics, EOL preferences, religion and spiritual beliefs were associated with use of moral terminology. RESULTS: We found two latent classes: a class in which patients did not use many expressions of morality in their PC consultations and one in which patients did. Age, race (white), education, spiritual needs, and whether a patient was affiliated with Christianity or another religion were all associated with membership of the first class. Gender, financial security and preference for longevity-focused over comfort focused treatment near EOL did not affect class membership. CONCLUSIONS: This study is among the first to use text data from a real-world situation to extract information regarding individual foundations of morality. It is the first to test empirically if individual moral expressions are associated with individual characteristics, attitudes and emotions.