Simple Baseline Research Articles

US-SOMO HPLC-SAXS module: dealing with capillary fouling and extraction of pure component patterns from poorly resolved SEC-SAXS data.

Size-exclusion chromatography coupled with SAXS (small-angle X-ray scattering), often performed using a flow-through capillary, should allow direct collection of monodisperse sample data. However, capillary fouling issues and non-baseline-resolved peaks can hamper its efficacy. The UltraScan solution modeler (US-SOMO) HPLC-SAXS (high-performance liquid chromatography coupled with SAXS) module provides a comprehensive framework to analyze such data, starting with a simple linear baseline correction and symmetrical Gaussian decomposition tools [Brookes, Pérez, Cardinali, Profumo, Vachette & Rocco (2013 ▸). J. Appl. Cryst.46, 1823-1833]. In addition to several new features, substantial improvements to both routines have now been implemented, comprising the evaluation of outcomes by advanced statistical tools. The novel integral baseline-correction procedure is based on the more sound assumption that the effect of capillary fouling on scattering increases monotonically with the intensity scattered by the material within the X-ray beam. Overlapping peaks, often skewed because of sample interaction with the column matrix, can now be accurately decomposed using non-symmetrical modified Gaussian functions. As an example, the case of a polydisperse solution of aldolase is analyzed: from heavily convoluted peaks, individual SAXS profiles of tetramers, octamers and dodecamers are extracted and reliably modeled.

Color image classification via quaternion principal component analysis network

The principal component analysis network (PCANet), which is one of the recently proposed deep learning architectures, achieves the state-of-the-art classification accuracy in various datasets and reveals a simple baseline for deep learning networks. However, the performance of PCANet may be degraded when dealing with color images due to the fact that the architecture of PCANet cannot properly utilize the spatial relationship between each color channel in three dimensional color image. In this paper, a quaternion principal component analysis network (QPCANet), which extends PCANet by using quaternion theory, is proposed for color image classification. Compared to PCANet, the proposed QPCANet takes into account the spatial distribution information of RGB channels in color images and ensures larger amount of intra-class invariance by using quaternion domain representation for color images. Experiments conducted on different color image datasets such as UC Merced Land Use, Georgia Tech face, CURet and Caltech-101 have revealed that the proposed QPCANet generally achieves higher classification accuracy than PCANet in color image classification task. The experimental results also verify that QPCANet has much better rotation invariance than PCANet when color image dataset contains lots of rotation information and demonstrate even a simple one-layer QPCANet may obtain satisfactory accuracy when compared with two-layer PCANet.

Efficient Utilization of Shared Caches in Multicore Architectures

Today most of the systems run general purpose applications which have non-uniform memory accesses. This work proposes a new technique for last level cache (LLC) organization, named as hashed cache using overuse distance for ways sharing (HCOWS), to reorganize the cache memory of a system to contribute better performance in multicore systems. The proposed HCOWS technique can be applied on multicore platforms, to improve performance of cache memory and hence of a system for myriad of general purpose applications. A new mapping technique is employed using universal hashing which is highly random. This scheme employs sharing of ways instead of set sharing. This work evaluates the performance of cache hierarchy using average memory access time. HCOWS scheme-based system has given better miss rate reduction when compared with the simple baseline system by 33 % in dual core, 37 % in quad core and 41 % in octa core, which substantiates the fact that increased number of cores dispenses better performance. The proposed scheme has high impact on miss rate reduction with the increase in size of LLC rather than the associativity of LLC.

A Predictive Model for Selecting Patients with HCV Genotype 3 Chronic Infection with a High Probability of Sustained Virological Response to Peginterferon Alfa-2a/Ribavirin.

BackgroundAccess to direct-acting antiviral agents (DAAs) is restricted in some settings; thus, the European Association for the Study of the Liver recommends dual peginterferon/ribavirin (PegIFN/RBV) therapy wherever DAAs are unavailable. HCV genotype (GT) 3 infection is now the most difficult genotype to eradicate and PegIFN/RBV remains an effective option. The goal of this study was to devise a simple predictive score to identify GT3 patients with a high probability of achieving a sustained virologic response (SVR) with PegIFN alfa-2a/RBV therapy.MethodsRelationships between baseline characteristics and SVR were explored by multiple logistic regression models and used to develop a simple scoring system to predict SVR using data from 1239 treatment-naive GT3 patients who received PegIFN alfa-2a/RBV for 24 weeks in two large observational cohort studies.ResultsThe score was validated using a database of 473 patients. Scores were assigned for six factors as follows: age (years) (≤40: 2 points; >40 but ≤55: 1); bodyweight (kg) (<70: 2; ≥70 but <90: 1); no cirrhosis/transition to cirrhosis (2); ALT ≤2.5 x ULN (1); platelets (109/L) (>200: 2; ≥100 but <200: 1); HCV RNA (<400,000 IU/mL: 1). The points are summed to arrive at a score ranging from 0‒10 where higher scores indicate higher chances of SVR; 141, 123, 203, 249, 232, and 218 patients had total scores of 0‒4, 5, 6, 7, 8, and 9–10, respectively, among whom SVR rates were 45%, 62%, 72%, 76%, 84%, and 89%. Among 622 patients who had scores of 6‒10 and HCV RNA <50 IU/mL by treatment week 4 the SVR rate was 86% (532/622).ConclusionsA simple baseline scoring system involving age, bodyweight, cirrhosis status, ALT level, platelet count and HCV RNA level can be used to identify treatment-naive Caucasian patients with HCV GT3 infection with a high probability of SVR with PegIFN alfa-2a/RBV therapy.

A POMDP Formulation of Proactive Learning

We cast the Proactive Learning (PAL) problem—Active Learning (AL) with multiple reluctant, fallible, cost-varying oracles—as a Partially Observable Markov Decision Process (POMDP). The agent selects an oracle at each time step to label a data point, while it maintains a belief over the true underlying correctness of its current dataset’s labels. The goal is to minimize labeling costs while considering the value of obtaining correct labels, thus maximizing final resultant classifier accuracy. We prove three properties that show our particular formulation leads to a structured and bounded-size set of belief points, enabling strong performance of point-based methods to solve the POMDP. Our method is compared with the original three algorithms proposed by Donmez and Carbonell and a simple baseline. We demonstrate that our approach matches or improves upon the original approach within five different oracle scenarios, each on two datasets. Finally, our algorithm provides a general, well-defined mathematical foundation to build upon.

Relational Knowledge Transfer for Zero-Shot Learning

General zero-shot learning (ZSL) approaches exploit transfer learning via semantic knowledge space. In this paper, we reveal a novel relational knowledge transfer (RKT) mechanism for ZSL, which is simple, generic and effective. RKT resolves the inherent semantic shift problem existing in ZSL through restoring the missing manifold structure of unseen categories via optimizing semantic mapping. It extracts the relational knowledge from data manifold structure in semantic knowledge space based on sparse coding theory. The extracted knowledge is then transferred backwards to generate virtual data for unseen categories in the feature space. On the one hand, the generalizing ability of the semantic mapping function can be enhanced with the added data. On the other hand, the mapping function for unseen categories can be learned directly from only these generated data, achieving inspiring performance. Incorporated with RKT, even simple baseline methods can achieve good results. Extensive experiments on three challenging datasets show prominent performance obtained by RKT, and we obtain 82.43% accuracy on the Animals with Attributes dataset.

Correct machine learning on protein sequences: a peer-reviewing perspective.

Machine learning methods are becoming increasingly popular to predict protein features from sequences. Machine learning in bioinformatics can be powerful but carries also the risk of introducing unexpected biases, which may lead to an overestimation of the performance. This article espouses a set of guidelines to allow both peer reviewers and authors to avoid common machine learning pitfalls. Understanding biology is necessary to produce useful data sets, which have to be large and diverse. Separating the training and test process is imperative to avoid over-selling method performance, which is also dependent on several hidden parameters. A novel predictor has always to be compared with several existing methods, including simple baseline strategies. Using the presented guidelines will help nonspecialists to appreciate the critical issues in machine learning.

Neer maneuver associated with a countertest in lateral rotation: A predictive test of response to dynamic humeral centering

Dynamic humeral centerering (DHC) is a physiotherapy modality that aims at preventing subacromial impingement. Clinical tests of subacromial impingement may be predictive of response to DHC in degenerative rotator cuff disorders (DRCD). The objective of the study was to assess the predictive value of Neer test and a countertest by passive antero-lateral elevation, the arm in lateral rotation (LR), with response to DHC as reference. Patients suffering from DRCD were prospectively included. Neer and RL tests were performed before RHD. Neer test was positive if painful (Neer + ). LR maneuver was positive if the pain disappeared during passive antero-lateral elevation with the arm in lateral rotation (RL + ). RHD was performed in ambulatory setting for 2 months. Response to RHD was expressed as percentage of variation and responder meaning variation >20%, using SPADI and Constant scores. 53 patients were included: age 59 ± 11 yr, ratio F/M 37/16, pain duration 27 ± 55 mo. Baseline clinical profiles were as follows: 36 patients Neer+, 17 Neer–, 20 LR+, 33 LR–. Neer test was not associated with outcomes after DHC. Conversely, LR+ patients had higher improvement than LR- patients (SPADI 43% vs.–15, P = 0,02; Constant 37 vs. 10, P = 0,01). LR+ patients were more frequently responders to RHD (SPADI 67% vs. 28, OR 5,1 [IC95: 1,5-17,8], P = 0,01; Constant 63 vs. 30, OR: 3,9 [1,2-13], P = 0,02). Baseline LR maneuver categorized responders and non-responders with following accuracy: 70% [IC95: 57-83] for SPADI score and 67% [54-80] for Constant score. This study demonstrates variability of response to DHC according to simple baseline clinical profiles. LR+ patients showed better response to DHC. LR maneuver may be a predictive test of response to RHD. It could be useful in therapeutic decision making in DRCD.

Predictors of treatment outcome in individuals with eating disorders: A systematic review and meta-analysis.

Understanding the factors that predict a favourable outcome following specialist treatment for an eating disorder may assist in improving treatment efficacy, and in developing novel interventions. This review and meta-analysis examined predictors of treatment outcome and drop-out. A literature search was conducted to identify research investigating predictors of outcome in individuals treated for an eating disorder. We organized predictors first by statistical type (simple, meditational, and moderational), and then by category. Average weighted mean effect sizes (r) were calculated for each category of predictor. The most robust predictor of outcome at both end of treatment (EoT) and follow-up was the meditational mechanism of greater symptom change early during treatment. Simple baseline predictors associated with better outcomes at both EoT and follow-up included higher BMI, fewer binge/purge behaviors, greater motivation to recover, lower depression, lower shape/weight concern, fewer comorbidities, better interpersonal functioning and fewer familial problems. Drop-out was predicted by more binge/purge behaviors and lower motivation to recover. For most predictors, there was large interstudy variability in effect sizes, and outcomes were operationalized in different ways. There were generally insufficient studies to allow analysis of predictors by eating disorder subtype or treatment type. To ensure that this area continues to develop with robust and clinically relevant findings, future studies should adopt a consistent definition of outcome and continue to examine complex multivariate predictor models. Growth in this area will allow for stronger conclusions to be drawn about the prediction of outcome for specific diagnoses and treatment types.

The prospects of easier security for small organisations and consumers

Safeguards exist that provide at least a reasonable degree of protection for IT security. With so much accumulated knowledge in existence, why aren't consumer devices delivered with convenient security facilities? This paper identifies the reasons why neither consumers nor providers of these technologies take responsibility for security. It presents a framework within which simple baseline security can be established for small organisations, and consumers can also achieve lower levels of insecurity than currently prevail. It then investigates the scope for interventions to achieve 'easy security' for small organisations and consumers.

Observation and prediction of three-dimensional morphology at a high-energy macrotidal beach

Three-dimensional beach features such as crescentic sandbars and rip channels influence beach response to, and recovery from, storm waves, as well as significantly affecting the safety and amenity provided by the surf-zone for beach water-users. In this contribution temporal variations in subtidal and intertidal beach three-dimensionality are observed at a high-energy macrotidal beach, and a simple equilibrium model is developed to predict the changes over multi-year timescales. A dataset of 5.5years of quasi-weekly bar measurements, and quasi-monthly intertidal surveys from Perranporth beach (Cornwall, UK) were used to quantify seasonal to inter-annual changes in three-dimensionality. The three-dimensionality of the outer bar displayed significant annual periodicity, with annual minima and maxima occurring in winter and spring, respectively. The lower intertidal beach displayed a similar periodicity, but developed three-dimensionality 1–4months before the outer bar. The model predicts increases or decreases in the scale of three-dimensional features by examining the disparity between instantaneous wave conditions and a temporally varying equilibrium wave condition. A tidally-modulated wave power term determines the rate of morphological change. Negative feedback was found to be an important process governing the changes in three-dimensionality; while free morphological behaviour may drive three-dimensional growth, negative feedback exerts stability in the system, making it inherently predictable using a temporally varying equilibrium value. The model explained 42% and 61% of the overall variability in outer bar and lower beach three-dimensionality, respectively. It skilfully predicted changes outside the training data range, during the most energetic 8-week period of waves measured in the last 65years off SW England, in winter 2013/14. The model outperformed a simple baseline model (a linear fit), as well as a comparable linearized feedback model from the literature, providing the first long-term (multi-year) predictions of seasonal to inter-annual beach three-dimensionality for a macrotidal beach.

Primary status, complementary status, and organizational survival in the U.S. venture capital industry

We introduce a distinction between two kinds of status and examine their effects on the exit rates of organizations investing in the U.S. venture capital industry. Extending past work on status-based competition, we start with a simple baseline: we describe primary status as a network-related signal of an organization’s quality in a leadership role, that is, as a function of the degree to which an organization leads others that are themselves well regarded as lead organizations in the context of investment syndicates. Combining Harary’s (1959) image of the elite consultant with Goffman’s (1956) concept of “capacity-esteem,” we then discuss complementary status as an affiliation-based signal of an organization’s quality in a supporting role. We measure complementary status as a function of the extent to which an organization is invited into syndicates by well-regarded lead organizations—that is, by those possessing high levels of primary status. Findings show that, conditioning on primary status, complementary status reduces the rate at which venture capital organizations exit the industry. Consistent with the premise that these kinds of status correspond to different roles and market identities, we also find that complementary status attenuates (and ultimately reverses) the otherwise favorable effect of primary status on an organization’s life chances. Theoretically and methodologically oriented scope conditions, as well as implications for future research, are discussed.

MetLab: an in silico experimental design, simulation and validation tool for viral metagenomics studies

The use of metagenomics for detection and characterisation of viruses has several difficult steps, including high throughput sequencing and bioinformatics analysis. We present MetLab, a new program aimed at providing scientists with a simple baseline for experimental design and analysis of viral metagenomics.

P1.07 - Outcome of patients with relapsed/refractory lymphoma in a large cohort inside a phase 1 clinic department

ABSTRACT Background: Treating relapsed/refractory lymphoma remains a challenge. While phase 1 trials classically aim to determine the recommended phase 2 dose (RP2D), the search of anti-tumor activity is increasingly evaluated. We aimed to assess the tolerance and efficacy of phase 1 trial and to determine a simple scoring system to identify patients who will prematurely discontinue phase 1 studies (before six weeks), in a large cohort of patients with relapsed/refractory lymphoma. Patients and Methods: Data from 105 consecutive patients with relapsed/refractory lymphoma treated within a panel of 17 phase 1 trials were collected, between 2008 and 2014. At inclusion, median age was 66 years [range: 23-83]. Lymphoma histological patient's types were: 58 (55%) aggressive non-Hodgkin lymphoma (34 diffuse large B-cell lymphoma, 7 T-cell lymphoma and 17 Mantle cell lymphoma), 31 (30%) indolent non-Hodgkin lymphoma and 16 (15%) Hodgkin lymphoma. The predefined Gustave Roussy (GR) score combined two simple variables, PS and baseline serum albumin (+1 if PS = 0, +1 if albumin ≤ 35g/l). Results: Grade 3 or 4 adverse events were experienced by 40/105 (38%) patients. With a median follow-up of 10 months, median OS and progression free survival (PFS) were respectively 19 (CI95%: 12-37) and 4 (CI95%: 2-5) months. Best overall response rate and disease control rate were 22% and 56%, respectively. Histological type's analysis shown median OS was 10, 45 and 47 months in aggressive-NHL, Hodgkin and indolent-NHL, respectively (p 0, baseline albumin ≤ 35 g/l and baseline LDH ≥ 250 UI/L were significantly associated with poorer OS. Patients with a GR score = 0 experienced significantly better OS compared to patients with a score = 1 and a score = 2 (37 months vs 17 months vs 9 months; p = 0.007). A premature study discontinuation was recorded in 28/105 patients (27%). The GR score distinguishes patients most likely to remain on study for more than 6 weeks. Conclusion: The three parameters WHO performance status (PS) > 0, baseline albumin ≤ 35 g/l and baseline LDH ≥ 250 UI/L are associated with OS and premature withdrew of study. Both PFS and OS survival curves demonstrates a plateau, in favor of therapeutic effect potentially maintained.

Predicting early mortality in resectable pancreatic adenocarcinoma: A cohort study.

Survival after surgical resection for pancreatic cancer remains poor. A subgroup of patients die early (<6 months), and understanding factors associated with early mortality may help to identify high-risk patients. The Khorana score has been shown to be associated with early mortality for patients with solid tumors. In the current study, the authors evaluated the role of this score and other prognostic variables in this setting. The current study was a cohort study of patients who underwent surgical resection for pancreatic cancer from January 2006 through June 2013. Baseline (diagnosis ±30 days) parameters were used to define patients as high risk (Khorana score ≥3). Statistically significant univariable associations and a priori prognostic variables were tested in multivariable models; adjusted hazard ratios (HR) were calculated. The study population comprised 334 patients. The median age was 67 years, 50% of the study population was female, and 86% of the patients were white. The pancreatic head was the primary tumor site for 73% of patients; 67% of tumors were T3 and 63% were N1. The median Khorana score was 2; 152 patients (47%) were determined to be high risk. Adjunctive treatment included chemotherapy (70%) and radiotherapy (40%). The postoperative (30-day) mortality rate was 0.9%. The 6-month mortality rate for the entire cohort was 9.4%, with significantly higher rates observed for high-risk patients (13.4% vs 5.6%; P = .02). On multivariable analyses (examining a total of 326 patients), the Khorana score (HR for high risk, 2.31; P = .039) and elevated blood urea nitrogen (HR, 4.34; P<.001) were associated with early mortality. Patients at high risk of early mortality after surgical resection of pancreatic adenocarcinoma can be identified using simple baseline clinical and laboratory parameters. Future studies should address preoperative interventions in these patients at high risk of early mortality.

Explicit Embeddings for Nearest Neighbor Search with Mercer Kernels

Many approximate nearest neighbor search algorithms operate under memory constraints, by computing short signatures for database vectors while roughly keeping the neighborhoods for the distance of interest. Encoding procedures designed for the Euclidean distance have attracted much attention in the last decade. In the case where the distance of interest is based on a Mercer kernel, we propose a simple, yet effective two-step encoding scheme: first, compute an explicit embedding to map the initial space into a Euclidean space; second, apply an encoding step designed to work with the Euclidean distance. Comparing this simple baseline with existing methods relying on implicit encoding, we demonstrate better search recall for similar code sizes with the chi-square kernel in databases comprised of visual descriptors, outperforming concurrent state-of-the-art techniques by a large margin.

Predicting early mortality in resectable pancreatic adenocarcinoma: A cohort study.

414 Background: Overall survival after pancreatic cancer resection remains poor. A subgroup of patients die early (&lt;6 months) and understanding factors associated with early mortality may help identify high-risk patients. The Khorana Score, including baseline hemoglobin, leukocyte and platelet counts, and body mass index [Khorana et al, Blood, 2008], has been shown to be associated with early mortality in solid tumors [Ay et al; Kuderer et al]. We evaluated the role of this score and other prognostic variables in predicting early mortality following resection. Methods: We conducted a cohort study of consecutive patients who underwent surgical resection for pancreatic cancer from January 2006 through June 2013 and were followed at the Cleveland Clinic. Baseline (diagnosis +/- 30 days) parameters were used to define patients as high-risk (Score &gt;=3). Statistically significant univariable associations and a priori prognostic variables were tested in multivariable models; adjusted hazard ratios (HR) are presented. Results: The study population comprised 334 patients. Median age was 67 years; 50% were female; 86% were Caucasian. Pancreatic head was the primary site for 73%; 67% were T3 and 63% were N1 tumors. Median Khorana score was 2; 152 patients (47%) were high-risk. Adjunctive treatment included chemotherapy (70%) and radiation (40%). Post-operative (30-day) mortality was 0.9%. Six-month mortality for the entire cohort was 9.4%, with significantly higher rates for high-risk patients (13.4% vs. 5.6%, p=0.02). In univariable analyses, Khorana score, low hemoglobin, and elevated blood urea nitrogen (BUN) were associated with early mortality; T- and N-stage, as well as margin status (R0/R1), were not. In multivariable analyses (n=326), Khorana score (HR high-risk = 2.31, p=0.039) and elevated BUN (HR = 4.34, p&lt;0.001) were associated with early mortality. Conclusions: Patients at high risk for early mortality after resection of pancreatic adenocarcinoma can be identified using simple baseline patient-related clinical and laboratory parameters rather than tumor characteristics. Future studies could evaluate the benefit of pre-operative interventions targeting patients at high risk for early mortality.

A Probabilistic Model Based Approach For Aligning Multiple Audio Sequences

We formulate the alignment problem of multiple and partially overlapping audio sequences in a probabilistic framework. We define and compare five generative models for several time varying features extracted from audio clips that are recorded independently and asynchronously. For each model, we derive the associated scoring function that evaluates the quality of an alignment. The matching is then achieved with a sequential algorithm. The derived score functions are also able to identify the cases where the sequences do not overlap and handle multiple sequences where no sequence is covering the entire timeline. The simulation results on real data suggest that the approach is able to handle difficult, ambiguous scenarios and partial matchings where simple baseline methods such as correlation fail.

A Review of Concussion Management in the Young Adult; the Importance of Baseline Testing and Caregiver Education

: Despite the increase in studies and media coverage, concussion continues to threaten the futures of young athletes and cause a great deal of concern among parents and practitioners. Questions continue regarding the management and return to play for injured adolescents and young adults, and regarding the long term consequences of repeated concussion in our youth. There appears to be a huge disparity between what health care professionals know and understand about concussion and what the average athlete and parent understands about what concussion is and the real risks involved. A review of current literature reveals a persistent lag between our knowledge and practice. This literature survey takes the athlete from the pre-participation sports physical, to the sidelines, to the doctor’s office, to return to play and provides a comprehensive assessment of current recommendations. There is a significant need, especially in urban areas, for education and access to cheap and simple baseline functional assessments of adolescents and young adults. We recommend athletes be assessed at their yearly physical in order to properly determine their baseline function and readiness to return to play after concussion. Additionally, we recommend providing anticipatory guidance and a simple concussion evaluation tool to be used by parents and guardians to also annually assess a young person’s baseline functional status and subsequent alterations. The goal of this review is to create an evidence-based, simple, cost-effective parental survey; increase awareness, understanding and diagnosis of concussion; and finally, expedited proper treatment and facilitate return to play. Publication indicesPubMed, lib.wayne.edu

A Simple Scoring System for Identifying Relapsed/Refractory Lymphoma Patients Prematurely Withdrawn from Phase I Trials: The Gustave Roussy Experience

Abstract Background: Treating relapsed/refractory non-Hodgkin and Hodgkin lymphoma in fit young and elderly patients represents a considerable challenge for clinicians. Combined cytotoxic agents are rapidly ineffective, especially when relapse occurs after autologous stem cell transplantation (ASCT). Alternative regimens are often sparse. In the era of personalized medicine, phase I clinical trials offer an alternative therapeutic option through a multitude of immunotherapy and targeted drugs in development. Phase I trials aim to determine the recommended phase II dose (RP2D) through toxicity and pharmacokinetic assessments. Documenting signal of activity is also a major objective, underlying the importance of including patients susceptible to remain on study until first lymphoma response evaluation (usually 6 weeks, 2 cycles). We aimed to assess a simple scoring system that could identify patients who will discontinue phase I studies before 6 weeks. Patients and Methods: Data from all lymphoma patients treated within a phase I trial in Gustave Roussy (GR) Cancer Center were retrospectively collected. 83 consecutive patients were enrolled in 17 phase I trials at GR between January 2008 and May 2014. All patients had progressive lymphoma at time of enrollment, after a median of 3 prior therapeutic lines (range 1;13). 37 patients (45%) received an ASCT prior to phase I inclusion. Median age was 67 years (range: 23-92) and WHO performance status (PS) was 0 in 36 patients (43%), 1 in 43 patients (52%) and 2 in 4 patients (5%). Median time from lymphoma diagnosis to phase I inclusion was 47 months. Lymphoma histological subtypes were represented as follows: 21% Hodgkin lymphoma, 36% aggressive non-Hodgkin lymphoma (83% diffuse large B cell lymphoma, 17% T cell lymphoma), 24% indolent non-Hodgkin lymphoma (70% follicular lymphoma, 25% marginal zone lymphoma, 5% Waldenström macroglobulinemia) and 19% mantle cell lymphoma. Univariate analysis on this cohort allowed identifying simple factors significantly associated with overall survival (OS). A simple scoring system, predictive for OS was pre-established and validated through a multivariate analysis in 2 large cohorts of various hematological malignancies by our group. Statistical tests were conducted on this relapsed/refractory lymphoma cohort, to evaluate this score's OS and early study discontinuation predictive ability. Results: Within a median follow up of 19 months, median OS and progression free survival (PFS) were respectively 18 (CI95%: 12; 37) and 3 (CI95%: 1.7; 3.6) months. Best overall response rate (ORR) and disease control rate (DCR: objective response and stable disease rates) were respectively 18% and 61%. Thirty-four patients (41%) experienced grade 3 or 4 adverse events and 7 patients (8%) a dose limiting toxicity (DLT). WHO performance status (PS) &gt; 0 at inclusion, baseline albumin ≤ 35 g/l and baseline LDH ≥ 250 UI/L were significantly associated with poorer OS. The predefined GR prognostic score combined 2 simple variables, PS and baseline serum albumin (+1 if PS 0, +1 if albumin≤ 35 g/l). In our lymphoma cohort, patients with a GR score = 0 experienced significantly better OS compared to patients with a score = 1 and a score = 2 (37 months vs 17 months vs 9 months; p = 0.007). This simple score, distinguishes patients most likely to remain on study for more than 6 weeks, considered as the minimum required period for response and toxicity evaluation (classical DLT and targeted therapies associated long term toxicities). Among patients prematurely withdrawn from clinical trials, 91% had a GR score ≥ 1 (p=0,007). Main study discontinuation reason was progressive disease (77%), drug related toxicity was only responsible for study discontinuation in 13 % of cases. Conclusion: Our data demonstrate efficacy and safety of phase I clinical trials in lymphoma, thus offering an interesting alternative therapeutic option for fit young and elderly relapsed/refractory lymphoma patients. The GR simple score, can both help in selecting patients most likely to benefit from a phase I trial (better OS) and to determine the phase II recommended dose (reduced early trial discontinuation). Disclosures No relevant conflicts of interest to declare.