This dissertation examines the effects of migration and remittances on family members left behind in sending areas. It consists of a systematic review of the literature on the effects of remittances on recipients’ health and empirical analyses that examine the effects of migration and remittances on medical spending and women’s empowerment. In 2019, 270 million international migrants sent over $530 billion dollars in remittances to their home countries. This project advances the literature on the effects of these processes. The empirical analyses use panel and cross-sectional survey data from the 2013 and 2018 waves of the Kerala Migration Survey, a large-scale longitudinal survey conducted in Kerala, India that followed the families of thousands of migrants from 1998 to 2018. The panel analyses used logistic and linear fixed-effects models, and the cross-sectional analyses used factor analysis, linear regression, and two-step Heckman selection models to assess the effects of migration and remittances on medical spending and women’s empowerment. The review found that migration and remittances play an important role in shaping health among individuals in transnational families. Remittances had the most pronounced benefits in areas with limited resources and social protections. They improved health by enabling households to pay for healthy food, medical treatment, and housing, and they improved mental health by easing financial constraints. Remittances were less beneficial in areas with strong safety nets, support systems, and community ties because family members were not dependent on remittances to provide basic needs. Migration and remittances shaped medical spending. Panel and cross-sectional analyses found that migrant households spent more money on medical expenses compared to non-migrant households. The panel analysis found that, compared to non-migrant households, households that received low levels of remittances spent less on medical care and households that received high levels of remittances spent a greater amount of money on medical care. After adjusting for health status and covariates, the cross-sectional analysis found positive and linear relationships between remittances and amount of medical spending. Although out-of-pocket costs of medical care in India can be high and although remittances are often used to maintain and increase social status, remittances were positively associated with increased medical spending in transnational families. Migration and remittances also influenced dimensions of women’s empowerment. They increased women’s management of household responsibilities, including decision-making and income control. However, migration decreased women’s internalized social status and neither migration nor remittances had meaningful effects on spousal attitudes. These findings suggest that women in migrant households may have taken on more household responsibilities, but their social status did not improve. This dissertation’s results suggest that migration and remittances influence health outcomes, certain dimensions of women’s empowerment, and ability to pay for to medical care in Kerala and other settings. However, their effects vary by context and indicators used to measure outcomes. Programs and policies should maximize the benefits of migration and remittances while reducing their associated harms.

This dissertation describes the patient experience. In Chapter One, we review the currentstate of patient-centered care with a specific focus on women’s experiences in infertility treatments, highlighting the gaps in our understanding of their experiences. The four independent but complementary aims of the dissertation studies are then introduced, and we identify how they address current gaps to advance our understanding of women’s experiences using reproductive technology. In Chapter One, we also introduce an innovative service design tool, journey mapping, as well as the Burden of Treatment Framework which guided the dissertation. Chapter Two, An Integrative Review of Journey Mapping to Document the Patient Experience, presents an integrative review examining the use of journey mapping in health care services research. In the analysis of these twenty-two studies, we demonstrate how journey mapping has been adapted to describe the patient experience. While the qualitative rigor of the included studies is of good quality, the inconsistent application of design standards in the accompanying visualizations, when present, suggests that further work and guidance is needed in the adaptation of this service design tool for the health research field. In Chapter Three, Journey Mapping the Patient Experience of IVF: A Social Media Analysis, we used posts from the largest infertility subreddit and patient-facing online resources to describe and visually depict the patient experience of IVF in two journey maps representing the experiences of women in their first cycle of infertility treatment and women in repeat cycles. Findings highlighted problems and unmet needs in the infertility treatment experience, including information needs, communication needs, and support needs. Chapter Four, The Burden of Treatment in IVF: An Analysis of Social Media Using a Framework for Chronic Complex Conditions, used qualitative descriptive methodology to guide inductive and deductive content analysis of posts from the same infertility subreddit. Findings suggest that Eton’s framework is applicable to infertility, but can be expanded by the inclusion of three new constructs that we identified. Finally, Chapter Five synthesizes the key findings across these four aims, outlining their strengths and limitations, and discussing implications for future research, policy, and clinical practice.

There is extensive research on the prevalence of college student engagement in risky sexual behaviors and its potentiating factors in college/university settings. However, there remains limited data and study on the use of campus health centers, as well as the patterns of how students seek sexual health care, its relationship to student gender identity, and student engagement in sexual risk behaviors. Despite literature surrounding student health centers and how students seek out medical care, the minimal data in the field regarding campus health center use, if and how students specifically seek out sexual health services, and how this collectively might vary across students with different gender identities, clearly indicate a need for further research on the experiences of students on college campuses in this regard. This dissertation is a secondary data analysis of the American College Health Association-National College Health Assessment III. This study’s specific aims were to: 1) describe key characteristics of students utilizing student health centers 2) describe if and how frequently college students are regularly seeking sexual health services; and 3) evaluate the relationship between a student’s gender identity and engagement in risky sexual behaviors. Drawing on a nationally representative sample of undergraduate students, analyses illustrated that students utilizing health services were primarily heterosexual female undergraduate students, with differing rates of utilization health services and sexual health services depending on race/ethnicity, year in school, and primary source of health insurance.The results confirmed that there are barriers to college student utilization of sexual health services, specifically for students who are non-heterosexual and transgender. Fewer students are seeking out their student health center and are seeking out care from services from other medical locations or may not be seeking out care at all. These findings highlight the need to ensure that student health centers on college campuses are inclusive of all students, regardless of gender identity. Given the important role that campus health center services play in promoting positive health outcomes and reducing sexual risk among its students, universities should consider efforts to increase student health center utilization, as it is a primary source of college student healthcare.

This study explored the intercultural conflict between urban Nepalis, who believe in Western medicine, and the Jhakris, Nepalese traditional healers. Nepalis in Western influenced urban areas made traditional healing practices illegal, resulting in violent arrests and beatings of Jhakris by police and occasional violent retaliation by Jhakris. This research sought to understand the roots of the intercultural conflict between the urban Nepali society and the Jhakris, and suggest possible solutions to reduce this intercultural conflict. Interviews were conducted with eight members of each group. The data were analyzed by using a conflict styles framework. Westernized members of urban Nepali society ranked high on the Competing conflict style and the Jhakri community ranked high in Compromise and Problem-Solving styles. The perceived grievances of each group were summarized and it was concluded that people in both groups who have a choice of Western and Jhakri healing options might be best placed to begin working toward negotiation and mediation.

In 2011 the National Academy of Medicine identified research gaps related to transgender populations and suggested a research agenda that included, among other goals, investigating health outcomes related to transition related care. The overarching goal of this dissertation therefore is to add to the body of knowledge about the state of health of transgender individuals, including utilization of gender-affirming care, preventive care and screening practices for human immunodeficiency virus (HIV) and other sexually transmitted infections (STIs). This dissertation includes three manuscripts. The first is a retrospective chart review including 1670 transgender patients, aged 18 and up (mean age 35.57 years), at a community health center to examine utilization of gender-affirming procedures as well as investigate the prevalence of smoking and uptake of colon cancer screening compared to New York City benchmarks using data from the New York City Community Health Survey (NYC CHS). The results revealed transgender individuals had high uptake of gender affirming hormones (81.9%) but fewer had undergone gender-affirming surgeries (31.5%). Transgender individuals had almost double the rate of current cigarette smoking compared to adults aged 18 and up in the New York City Community Health Survey (OR=1.92, 95% CI=1.61, 2.28) and also had suboptimal colon cancer screening rates compared to New Yorkers aged 50 and older (OR=0.16, 95% CI=0.11, 0.23). The second paper is a scoping review of the literature to investigate postoperative outcomes related to vaginoplasty procedures in transgender women. One hundred and three articles met inclusion criteria and provided information on immediate as well as long term health outcomes. The review demonstrated many inconsistencies in the timing of follow-up as well as how outcomes were measured, but provided invaluable information on the many types of postoperative issues that may be seen after vaginoplasty surgery. Lastly, the third paper examined the prevalence of HIV and STI testing behavior and prevalence of HIV infection among transgender people in a community health center setting. This analysis demonstrated that HIV screening rates were lower than expected (55.7%) given the elevated HIV prevalence in the population. In the multivariate analysis the odds of HIV screening among transmasculine individuals was higher in those who had undergone gender affirming surgeries (OR=1.67, 95% CI= 1.08, 2.58), had a substance use history (OR=5.18, 95% CI=1.41, 18.99) and a history of genital warts (OR=4.64, 95%CI=1.24, 17.34). Among transfeminine individuals the odds of HIV screening were higher in those with only cisgender male partners (OR=2.18, 95% CI=1.52, 3.11), gender affirming surgery (OR=2.56, 95% CI=1.53, 4.31), substance use history (OR=2.76, 95% CI=1.23, 5.78) and genital warts (OR=2.69, 95% CI=1.20, 6.02). HIV prevalence was higher among transfeminine compared to transmasculine individuals (28.1% vs. 2.8%, p<.001). In the multivariable analysis having only cisgender male sex partners increased the odds of HIV infection among transmasculine individuals (OR=10.58, 95% CI=1.33, 84.17), while having at least a high school diploma reduced the odds of infection (OR=0.08, 95% CI=0.01, 0.72). Among transfeminine individuals increased odds of HIV-infection were seen in those who were unemployed (OR=1.7, 95% CI=1.1, 2.64) and those who had a history of genital warts (OR=2.54, 95% CI=1.37, 4.70). White individuals had a lower likelihood of HIV infection (OR=0.40, 95%CI=0.21, 0.73). Overall these three studies provide important information about transition-related, primary and preventive healthcare for transgender populations. The findings of elevated cigarette smoking, underutilization of colorectal cancer screening and low HIV and STI screening rates occurred in this study despite the fact that transgender people were engaged in medical care. Clinics and other health settings that provide transgender health services should include robust metrics for monitoring uptake of preventive health care services and work to improve uptake of services when disparities are evident.

Emergency medical care (EMC) is proposed by the World Health Organization (WHO) as being one of the core components of a horizontal approach to improving population health in low-resource settings; triage is considered to be a fundamental part of this field. Most studies exploring triage have focused on high-income countries. In 2004, the Cape Triage Group (CTG) developed the South African Triage Scale (SATS) a scale that uses a physiologically based scoring system together with a list of discriminators - designed to triage patients into one of four priority groups for medical attention. The SATS was designed for use in the South African context to mitigate the limited numbers of doctors and professional nurses. The SATS has been implemented and assessed extensively in South Africa, but its performance across a spectrum of different low-resource settings, particularly non-sub-Saharan African and trauma-only settings, has not been adequately assessed. Médecins Sans Frontières (MSF), an international humanitarian organisation, introduced EMC in 2006 into low-resource settings. In 2011, MSF began introducing the SATS in various projects where it was providing EMC. Methodology: This was a multi-site retrospective cohort study which sought to assess the reliability and validity of the SATS in different low-resource settings. Aim 1: To implement and evaluate the SATS in Northern Pakistan by describing the steps of implementation and how accurate nurses were in using the triage scale. After one month of implementation, 370 triage forms from a one-week period were evaluated. Aim 2: To assess the inter- and intra-rater reliability and accuracy of nurse triage ratings when using the SATS in an emergency centre (EC) in Timergara, Pakistan. Fifteen EC nurses assigned triage ratings to a set of 42 reference vignettes (written case reports of EC patients) under classroom conditions. Inter-rater reliability was assessed by comparing these triage ratings; intra-rater reliability was assessed by asking the nurses to re-triage ten 12 random vignettes from the original set of 42 vignettes and comparing the duplicate ratings. Accuracy of the nurse ratings was measured against the reference standard. Aim 3: To improve the ability to measure reliability and validity in paediatric settings by developing a set of paediatric paper-based vignettes using the Delphi methodology. In a two-round consensus building process, a panel of EC experts were asked to independently triage 50 clinical vignettes using one of four acuity levels: emergency (patient to be seen immediately), very urgent (patient to be seen within 10 min), urgent (patient to be seen within 60 min), or routine (patient to be seen within four hours). The vignettes were based on real paediatric EC cases in South Africa. Vignettes that reached a minimum of 80% group consensus for acuity ratings on either round one or two were included in the final set of reference vignettes. Aim 4: To further assess the reliability of the SATS across MSF-supported hospitals using paper-based vignettes in Afghanistan, Haiti and Sierra Leone. Applying the same methodology as in Northern Pakistan, we assessed reliability under classroom conditions between December 2013 and February 2014. Aim 5: To assess the validity of the SATS across MSF-supported hospitals between June 2013 and June 2014. Validity was assessed by comparing patients’ SATS ratings with their final EC outcomes (i.e., hospital admission, death or discharge) across four sites in Afghanistan, Haiti and Sierra Leone. Findings The SATS was able to be easily implemented and accurately completed in a low-resource setting of Northern Pakistan. We recommended further implementation and assessment of reliability and validity in low-resource settings. Across six sites with a total of 87 nurses, including two trauma-only hospitals in Afghanistan and Haiti, a paediatric-only hospital in Sierra Leone and three mixed medical settings in Afghanistan, Pakistan and Haiti, the SATS demonstrated moderate to substantial reliability. Across all settings in which we measured validity using outcome markers, SATS predicted an increase in the likelihood of admission/death when moving from low- to high-triage acuity. In trauma-only settings of Afghanistan and Haiti, the SATS showed a 1-9% under-triage and 13 a 2-16% over-triage rate. In mixed medical and paediatric settings, under-triage ranged from 0-76% while over-triage ranged from 2-88%. A more logical standardised approach to assessing validity was put forward when using outcome markers that would allow easier comparisons to be done across validity studies irrespective of the number of levels the triage scale had. We developed a set of paediatric vignettes for use in low-resource settings but cautioned against its use after measuring reliability using adult reference vignettes. We found that generic vignettes were poor substitutes in a variety of settings based on a lack of contextualisation and understanding by local nurses. Conclusion: The SATS has reasonable reliability with good validity across different ECs in various lower-source settings. The SATS is a valid triage tool for prioritisation of patients with trauma in low-resource settings. Its use in mixed EC settings seems justified, but in paediatric settings context-specific adjustments and assessments of its performance would be prudent.

As the delivery and management of healthcare continues to become increasingly complex, confronted by numerous economic, political and public health challenges, it is crucial for healthcare organizations to effectively develop and support their clinical and administrative leaders. A national staffing shortage in healthcare coupled with a global pandemic has increased burnout for healthcare providers. Retaining, engaging, and developing talent has taken on an urgency not seen before in healthcare organizations. The academic healthcare system has additional layers of intricacy with its tripartite clinical, educational, and research missions. Collaboration – and the behaviors and attitudes that enable it – is key for all leaders to negotiate across disciplinary boundaries within the academic health enterprise. Healthcare organizations must carefully consider how to develop and empower leaders capable of both leading teams and leading collaboratively, with peer leaders across the system. This study used a mixed-method case study approach to evaluate an interprofessional, team-based leadership development program in an academic health system to identify the program elements and learning experiences that influenced the participants’ ability to demonstrate the tenets of collaborative leadership and the factors that fostered or inhibited their abilities. Mid-level level leadership teams were nominated to the 10-month program, which utilized a blended learning approach of in-person days, live virtual sessions and online content and interaction. Teams were also required to complete a capstone project to allow the opportunity for integrating program learning through action while contributing to organizational change and innovation. Historical data from focus groups, evaluations and pre and post self-efficacy assessments from six cohorts were analyzed for this study. Ten semi-structured interviews comprised the new data collected. Focus group and evaluation data was derived from all program participants from six cohorts, consisting of 145 participants, in which 97/145 (67%) were females and 48/145 (33%) were males. Physicians comprised 28% of the group, nurses/other clinical specialists 31%, and 40% were administrators. The pre and post self-efficacy assessment data was limited to 65 participants. It is evident from the findings that leadership development as an interprofessional team, and as part of a larger cohort, is a powerful way to improve team cohesion, build relationships across the healthcare system, and enable boundary crossing – all vital capabilities to collaborative leadership. Learning as a team revised participants’ mental models by enabling an appreciation of diverse perspectives and self-awareness. The design of the program, particularly the program’s length and the action learning experience, were key factors in allowing the leadership ability of teams to mature. The program gave participants the dedicated time and space to build trust, solve problems, and reflect as individuals and as team. During the program and after, the most prevalent challenge for leaders was finding the time to establish and maintain relationships, which is critical to collaborative leadership. Other limiting factors were leadership turnover, lack of senior leadership support, and barriers to physician engagement. The organization in this study, like other academic health systems, employs a matrixed structure. While the matrixed structure necessitates collaboration, inherent stressors such as ambiguity around roles, decision-making abilities, power, and competing priorities emerged as a challenge to collaborating with peer leaders. The major recommendations for leadership development in an academic health system derived from this study are to: 1) design leadership development learning experiences within the context of an interprofessional team, 2) consider how participants may be enabled and inhibited in their leadership practice by organizational structures and cultures and, 3) ensure that the program length allows for the adequate time and space to learn, integrate and apply new ways of leading.

El texto completo de este trabajo no está disponible en el Repositorio Académico UPC por restricciones de la casa editorial donde ha sido publicado. / In this article, we propose a technological architecture that supports the remote medical care service for elderly people suffering from Type 2 Diabetes Mellitus in Peru. The problem is based on the limitations that elderly people have with compliance with medical controls, which leads to increased complications of the disease and the patient's quality of life. The design of the technological architecture is based on 6 layers: 1) devices, 2) software, 3) channels, 4) data storage, 5) data processing and 6) information visualization. Through the solution, patients are able to autonomously manage their disease through periodic glucose control and the execution of an updated treatment in real time. In the same way, the specialist doctor periodically analyzes the glucose level and it is notified in real time about the out-of-range indicators, which allows him to make treatment decisions as the anomaly occurs. The focus of the solution is to reduce the complications generated by the disease through efficient glucose control and periodic medical advice. The solution was validated in a nursing home with adults over 60 and an endocrinologist from a medical office in Lima, Peru. For the study, we measured the constancy of the glucose record, the average response time of the doctor in case of emergencies or indicators out of range, the percentage of reduction of complications and the level of satisfaction of the Telehomecare solution in older adults. The results show that patients interact more frequently as they adapt the solution as part of their daily routine. On the other hand, the response time was approximately 4.13 minutes from the anomaly record. The percentage of complication reduction was 14% and the level of satisfaction of the solution was reflected in the dimensions of the response time and understanding of the user's need. / Revisión por pares

Heterogeneous treatment responses are commonly observed in patients with mental disorders. Thus, a universal treatment strategy may not be adequate, and tailored treatments adapted to individual characteristics could improve treatment responses. The theme of the dissertation is to develop statistical and machine learning methods to address patients heterogeneity and derive robust and generalizable individualized treatment strategies by integrating evidence from multi-domain data and multiple studies to achieve precision medicine. Unique challenges arising from the research of mental disorders need to be addressed in order to facilitate personalized medical decision-making in clinical practice. This dissertation contains four projects to achieve these goals while addressing the challenges: (i) a statistical method to learn dynamic treatment regimes (DTRs) by synthesizing independent trials over different stages when sequential randomization data is not available; (ii) a statistical method to learn optimal individualized treatment rules (ITRs) for mental disorders by modeling patients' latent mental states using probabilistic generative models; (iii) an integrative learning algorithm to incorporate multi-domain and multi-treatment-phase measures for optimizing individualized treatments; (iv) a statistical machine learning method to optimize ITRs that can benefit subjects in a target population for mental disorders with improved learning efficiency and generalizability. DTRs adaptively prescribe treatments based on patients' intermediate responses and evolving health status over multiple treatment stages. Data from sequential multiple assignment randomization trials (SMARTs) are recommended to be used for learning DTRs. However, due to the re-randomization of the same patients over multiple treatment stages and a prolonged follow-up period, SMARTs are often difficult to implement and costly to manage, and patient adherence is always a concern in practice. To lessen such practical challenges, in the first part of the dissertation, we propose an alternative approach to learn optimal DTRs by synthesizing independent trials over different stages without using data from SMARTs. Specifically, at each stage, data from a single randomized trial along with patients' natural medical history and health status in previous stages are used. We use a backward learning method to estimate optimal treatment decisions at a particular stage, where patients' future optimal outcome increment is estimated using data observed from independent trials with future stages' information. Under some conditions, we show that the proposed method yields consistent estimation of the optimal DTRs, and we obtain the same learning rates as those from SMARTs. We conduct simulation studies to demonstrate the advantage of the proposed method. Finally, we learn DTRs for treating major depressive disorder (MDD) by stage-wise synthesis of two randomized trials. We perform a validation study on independent subjects and show that the synthesized DTRs lead to the greatest MDD symptom reduction compared to alternative methods. The second part of the dissertation focuses on optimizing individualized treatments for mental disorders. Due to disease complexity, substantial diversity in patients' symptomatology within the same diagnostic category is widely observed. Leveraging the measurement model theory in psychiatry and psychology, we learn patient's intrinsic latent mental status from psychological or clinical symptoms under a probabilistic generative model, restricted Boltzmann machine (RBM), through which patients' heterogeneous symptoms are represented using an economic number of latent variables and yet remains flexible. These latent mental states serve as a better characterization of the underlying disorder status than a simple summary score of the symptoms. They also serve as more reliable and representative features to differentiate treatment responses. We then optimize a value function defined by the latent states after treatment by exploiting a transformation of the observed symptoms based on the RBM without modeling the relationship between the latent mental states before and after treatment. The optimal treatment rules are derived using a weighted large margin classifier. We derive the convergence rate of the proposed estimator under the latent models. Simulation studies are conducted to test the performance of the proposed method. Finally, we apply the developed method to real-world studies. We demonstrate the utility and advantage of our method in tailoring treatments for patients with major depression and identify patient subgroups informative for treatment recommendations. In the third part of the dissertation, based on the general framework introduced in the previous part, we propose an integrated learning algorithm that can simultaneously learn patients' underlying mental states and recommend optimal treatments for each individual with improved learning efficiency. It allows incorporation of both the pre- and post-treatment outcomes in learning the invariant latent structure and allows integration of outcome measures from different domains to characterize patients' mental health more comprehensively. A multi-layer neural network is used to allow complex treatment effect heterogeneity. Optimal treatment policy can be inferred for future patients by comparing their potential mental states under different treatments given the observed multi-domain pre-treatment measurements. Experiments on simulated data and real-world clinical trial data show that the learned treatment polices compare favorably to alternative methods on heterogeneous treatment effects and have broad utilities which lead to better patient outcomes on multiple domains. The fourth part of the dissertation aims to infer optimal treatments of mental disorders for a target population considering the potential distribution disparities between the patient data in a study we collect and the target population of interest. To achieve that, we propose a learning approach that connects measurement theory, efficient weighting procedure, and flexible neural network architecture through latent variables. In our method, patients' underlying mental states are represented by a reduced number of latent state variables allowing for incorporating domain knowledge, and the invariant latent structure is preserved for interpretability and validity. Subject-specific weights to balance population differences are constructed using these compact latent variables, which capture the major variations and facilitate the weighting procedure due to the reduced dimensionality. Data from multiple studies can be integrated to learn the latent structure to improve learning efficiency and generalizability. Extensive simulation studies demonstrate consistent superiority of the proposed method and the weighting scheme to alternative methods when applying to the target population. Application of our method to real-world studies is conducted to recommend treatments to patients with major depressive disorder and has shown a broader utility of the ITRs learned from the proposed method in improving the mental states of patients in the target population.

Markov Decision Processes (MDP) is a common framework for modeling sequential decision-making problems, with applications ranging from inventory and supply chains to healthcare applications, autonomous driving and solving repeated games. Despite its modeling power, two fundamental challenges arise when using the MDP framework in real-worldapplications. First, the optimal decision rule may be highly dependent on the MDP parameters (e.g., transition rates across states and rewards for each state-action pair). When the parameters are miss-estimated, the resulting decision rule may be suboptimal when deployed in practice. Additionally, the optimal policies computed by state-of-the-art algorithms may not be interpretable and can be seen as a black-box. Among other reasons, this is problematic as the policy may not be understandable for the people that are supposed to operationalize it. In this thesis, we aim to broaden the applicability of the MDP framework by addressing the challenges of robustness and interpretability. In the first part of the thesis, we focus on robustness. We introduce a novel model for parameter uncertainty in Chapter 2, that is significantly less pessimistic than prior models of uncertainty while enabling the efficient computation of a robust policy. In Chapter 3, we consider a healthcare application, where we focus on proactively transferring patients of a hospital to the Intensive Care Unit, to ameliorate the overall survival rates and patients’ flow. In the second part of this thesis, we focus on interpretable algorithms, with an emphasis on the application to find novel triage protocols for ventilator allocations for COVID-19 patients. In Chapter 4, we introduce a simulation model to estimate the performance of the official New York State (NYS) triage protocol at various levels of shortages of ventilators, using a real data set of patients intubated during Spring 2020 because of COVID-19 complications. In Chapter 5, we introduce our algorithmic framework for computing interpretable (tree) policies and apply our methods to learn novel triage protocols.