Search

The global cost of drug discovery and development exceeds $200 billion annually. The main results of drug discovery and development are the outcomes of clinical trials, which directly influence the regulatory approval of new drug candidates and ultimately affect patient outcomes. Despite their significance, large-scale, high-quality clinical trial outcome data are not readily available to the public. Suppose a large clinical trial outcome dataset is provided; machine learning researchers can potentially develop accurate prediction models using past trials and outcome labels, which could help prioritize and optimize therapeutic programs, ultimately benefiting patients. This paper introduces Clinical Trial Outcome (CTO) dataset, the largest trial outcome dataset with around 479K clinical trials, aggregating outcomes from multiple sources of weakly supervised labels, minimizing the noise from individual sources, and eliminating the need for human annotation. These sources include large language model (LLM) decisions on trial-related documents, news headline sentiments, stock prices of trial sponsors, trial linkages across phases, and other signals such as patient dropout rates and adverse events. CTO's labels show unprecedented agreement with supervised clinical trial outcome labels from test split of the supervised TOP dataset, with a 91 F1.

At the far reaches of the outer Solar System, the ice giants remain the last class of planets yet to be studied using orbiters. The 2023-2032 Planetary Science Decadal Survey has underscored the importance of the ice giants in understanding the origin, formation, and evolution of our Solar System. The enormous heliocentric distance of Uranus presents considerable mission design challenges, the most important being able to reach Uranus within a reasonable time. The present study presents two examples of aerocapture enabled short flight time, fast trajectories for Uranus orbiter missions, and highlights the enormous benefits provided by aerocapture. The first is an EEJU trajectory with a launch opportunity in July 2031 with a flight time of 8 years. The second is an EJU trajectory with a launch opportunity in June 2034 with a flight time of only 5 years. Using the Falcon Heavy Expendable, the available launch capability is 4950 kg and 1400 kg respectively for the two trajectories. Both trajectories have a high arrival speed of 20 km/s, which provides sufficient corridor width for aerocapture. Compared to propulsive insertion architectures which take 13 to 15 years, the fast trajectories offer significant reduction in the flight time., Comment: 9 pages, 6 figures

Aerocapture uses atmospheric drag to decelerate spacecraft and achieve orbit insertion. One of the significant risks associated with aerocapture is the uncertainty in the atmospheric density, particularly for outer planets. The paper performs a comparative study of the atmospheric uncertainties and provides design rules for aerocapture missions. The atmospheres of Venus, Mars, and Titan are well-characterized for engineering purposes. At the altitude ranges relevant for aerocapture, the 3$\sigma$ density variation is approximately $\pm$30%, $\pm$50%, $\pm$30% for Venus, Mars, and Titan respectively. With no in-situ data, the atmospheres of Uranus and Neptune are not as well characterized as the other bodies. For both Uranus and Neptune, the GRAM suite provides a 3$\sigma$ density variation of approximately $\pm$30% for the relevant altitude ranges which is considered an optimistic estimate. Until in-situ data from an atmospheric probe becomes available, a more conservative global min-max estimate is recommended to accommodate the worst-case scenario. The study presents a graphical method for selection of the optimal entry flight path angle when considering the atmospheric uncertainties to ensure the on-board guidance is given the best possible initial state for targeting the desired exit state post aerocapture., Comment: 14 pages, 10 figures

The choice of the launch vehicle is an important consideration during the preliminary planning of interplanetary missions. The launch vehicle must be highly reliable, capable of imparting sufficient energy to the spacecraft to inject it on to an Earth-escape trajectory, and must fit within the cost constraints of the mission. Over the recent past, the most commonly used launchers for interplanetary missions include the Atlas V401, Atlas V551, Delta IVH, and Falcon Heavy expendable version. The NASA Launch Vehicle Performance website maintains a tool to help mission planners evaluate various launch vehicles during mission studies. However, there is no comprehensive dataset which can be used to quickly compare the launch performance and launch cost of various options. The present study compiles a dataset of the high energy performance of existing and planned launchers from open-source data and performs a quantitative comparison of the launch performance and the launch cost per kg. The Falcon Heavy expendable offers the lowest cost-per-kg for high-energy launches, with only $0.075M per kg. The Vulcan Centaur offers comparable performance to the Falcon Heavy. The results indicate Falcon Heavy Expendable and the Vulcan Centaur will be the likely choice for several future missions., Comment: 6 pages, 4 figures

With advancements in low-cost launchers and small interplanetary spacecraft, NASA has recognized the potential of small missions to perform focused planetary science investigations at Mars and Venus. The EscaPADE, part of the NASA SIMPLEx program will deliver two small spacecraft to elliptical orbits around Mars using the Photon spacecraft. Orbit insertion, particularly to low-circular orbits requires significant propellant, taking up a substantial fraction of the Photon wet mass and present a significant challenge for small missions. The large $\Delta$V requirements for low-circular orbit make it difficult to insert small satellites into these orbits even with the highly capable Photon, as the total $\Delta$V for Earth escape and orbit insertion exceeds its capability. Drag modulation aerocapture offers a promising alternative, using the atmospheric drag to obtain the large $\Delta$V. The study shows how the Photon when combined with drag modulation aerocapture can deliver small orbiters to low-circular orbits, enabling a wide range of small orbiter missions. Aerocapture eliminates the need for Photon to provide 2 to 3.5 km/s of $\Delta$V for orbit insertion, which translate into mass and cost savings, and can enable frequent low-cost small orbiters and small satellite constellations at Mars and Venus in the near future., Comment: 9 pages, 4 figures

Aerocapture is an orbit insertion technique which uses atmospheric drag from a single pass to decelerate a spacecraft. Compared to conventional propulsive insertion, aerocapture can impart large velocity changes to the spacecraft with almost no propellant. At the far reaches of the outer Solar System, the ice giants remain the last class of planets to be explored using orbiters. Their enormous heliocentric distance presents significant mission design challenges, particularly the large $\Delta$V required for orbit insertion. This makes aerocapture an attractive method of orbit insertion, but also challenging due to the comparatively large navigation and atmospheric uncertainties. The present study performs a comparison of the lift and drag modulation control and their implications for future missions. Lift modulation provides nearly twice the entry corridor width as drag modulation, and can thus accommodate larger uncertainties. Lift modulation offers continuous control throughout the flight enabling it to adjust the trajectory in response to the actual density profile encountered. Drag modulation offers much more benign aero-thermal conditions compared to lift modulation. With drag modulation, there is no control authority after the drag skirt jettison making the vehicle more susceptible to exit state errors from density variations encountered after the jettison event., Comment: 7 pages, 3 figures

Aerocapture is a maneuver which uses aerodynamic drag to slow down a spacecraft in a single pass through the atmosphere. All planetary orbiters to date have used propulsive orbit insertion. Aerocapture is a promising alternative, especially for small satellite missions and missions to the ice giants. The large {\Delta}V requirement makes it practically impossible for small satellites to enter low circular orbits. Aerocapture can enable insertion of low-cost satellites into circular orbits around Mars and Venus. For ice giant missions, aerocapture can enable orbit insertion from fast arrival trajectories which are impractical with propulsive insertion. By utilizing the atmospheric drag to impart the {\Delta}V, aerocapture can offer significant propellant mass and cost savings for a wide range of planetary missions. The present study analyzes the performance benefit offered by aerocapture for a set of design reference missions and their applications to future Solar System exploration from Venus to Neptune. The estimated performance benefit for aerocapture in terms of delivered mass increase are: Venus (92%), Earth (108%), Mars (17%), and Titan (614%), Uranus (35%), and Neptune (43%). At Uranus and Neptune, aerocapture is a mission enabling technology for orbit insertion from fast arrival interplanetary trajectories., Comment: 11 pages, 8 figures

Deep neural network models can learn clinically relevant features from millions of histopathology images. However generating high-quality annotations to train such models for each hospital, each cancer type, and each diagnostic task is prohibitively laborious. On the other hand, terabytes of training data -- while lacking reliable annotations -- are readily available in the public domain in some cases. In this work, we explore how these large datasets can be consciously utilized to pre-train deep networks to encode informative representations. We then fine-tune our pre-trained models on a fraction of annotated training data to perform specific downstream tasks. We show that our approach can reach the state-of-the-art (SOTA) for patch-level classification with only 1-10% randomly selected annotations compared to other SOTA approaches. Moreover, we propose an uncertainty-aware loss function, to quantify the model confidence during inference. Quantified uncertainty helps experts select the best instances to label for further training. Our uncertainty-aware labeling reaches the SOTA with significantly fewer annotations compared to random labeling. Last, we demonstrate how our pre-trained encoders can surpass current SOTA for whole-slide image classification with weak supervision. Our work lays the foundation for data and task-agnostic pre-trained deep networks with quantified uncertainty., Comment: 18 pages, 8 figures

Thermal protection systems are a critical component of planetary exploration, enabling probes to enter the atmosphere and perform in-situ measurements. The aero-thermal conditions encountered during entry are destination and vehicle dependent, ranging from relatively benign conditions at Mars and Titan, to extreme conditions at Venus and Jupiter. The thermal protection system is a single-point-of-failure for both entry probe and aerocapture missions, and hence must be qualified using ground tests to ensure mission success. The high density Carbon-Phenolic which was used in the Galileo and the Pioneer Venus missions is no longer available due to the lack of the manufacturing base for its raw materials. To address the need for Venus and outer planet missions, NASA has developed the Heatshield for Extreme Environment Entry Technology (HEEET). The present study uses the Aerocapture Mission Analysis Tool (AMAT) to perform a comparative study of the thermal protection system requirements for various planetary destinations and the applicability of HEEET for future entry and aerocapture missions. The heat rate and stagnation pressure for aerocapture is significantly less compared to probe entry. The large heat loads during aerocapture present a challenge, but HEEET is capable of sustaining large heat loads within a reasonable TPS mass fraction., Comment: 14 pages, 12 figures

Aerocapture is the technique of using planetary atmospheres to decelerate a spacecraft in a single pass to achieve nearly fuel-free orbit insertion. Aerocapture has been extensively studied since the 1980s but has never been flown yet. The entry conditions encountered during aerocapture are strongly destination dependent, and performance benefit offered by aerocapture is also destination dependent. Aerocapture is applicable to all atmosphere-bearing destinations with the exception of Jupiter and Saturn, whose extreme entry conditions make aerocapture infeasible. A recent study by the NASA Science Mission Directorate highlighted the need for baseline design reference missions, as a starting point for system level architecture studies. The present study uses the Aerocapture Mission Analysis Tool (AMAT) to compile a list of design reference missions at Venus, Earth, Mars, Titan, Uranus, and Neptune. These reference missions can provide an initial assessment of the feasibility of aerocapture for a proposed mission, and provide intial baseline values for more detailed system studies. The reference mission set provides a quick estimate of the entry conditions, control requirements, and aero-thermal loads for architectural level studies., Comment: 12 pages, 7 figures

Since the beginning of robotic interplanetary exploration nearly six decades ago, successful atmospheric entry has been accomplished at Venus, Earth, Mars, Jupiter, and Titan. More entry probe missions are planned to Venus, Titan, and Uranus in the next decade. Atmospheric entry subjects the vehicle to rapid deceleration and aerothermal loads which the vehicle must be designed for, to deliver the robotic instruments inside the atmosphere. The design of planetary probes and their mission architecture is complex, and involves various engineering constraints such as peak deceleration, heating rate, heating load, and communications which must be satisfied within the budget and schedule of cost constrained mission opportunities. Engineering design data from previous entry probe missions serve as a valuable reference for designing future missions. The present study compiles an augmented version of the blue book entry probe dataset, performs a comparative analysis of the entry conditions, and provides engineering rules of thumb for design of future missions. Using the dataset, the present study proposes a new empirical correlation which aims to more accurately predict the thermal protection system mass fraction for high heat load conditions during entry and aerocapture at Uranus and Neptune., Comment: 15 pages, 15 figures

Over the history of planetary exploration, atmospheric entry vehicles have been used to deliver probes and landers to Venus, Mars, Jupiter, and Titan. While the entry vehicles are tools for furthering scientific exploration, by delivering probes and landers which perform in-situ exploration, the entry vehicle and trajectory design in itself is of significant interest. Entering an atmosphere subjects the vehicle to deceleration and aerodynamic heating loads which the vehicle must withstand to deliver the probe or the lander inside the atmosphere. The conditions encountered depend on the destination, the atmosphere-relative entry speed, the vehicle type, ballistic coefficient, the vehicle geometry, and the entry-flight path angle. The driving constraints are the peak aerodynamic deceleration, peak heat rate, and the total heat load encountered during the critical phase of the entry. Jupiter presents the most extreme entry conditions, while Titan presents the most benign entry environment. This study presents a survey of the design trade space for atmospheric entry missions across the Solar System using carpet plots, along with benchmarks from historical missions, which can serve as a useful reference for designing future missions., Comment: 11 pages, 8 figures

The study of planetary atmospheres is critical to our understanding of the origin and evolution of the Solar System. The combined effect of various physical and chemical processes over billions of years have resulted in a variety of planetary atmospheres across the Solar System. This paper performs a comparative study of planetary atmospheres and their engineering implications for future entry and aerocapture missions. The thick Venusian atmosphere results in high deceleration and heating rates and presents a demanding environment for both atmospheric entry and aerocapture. The thin Martian atmosphere allows low aerodynamic heating, but itself is not enough to decelerate a lander to sufficiently low speeds for a soft landing. With their enormous gravity wells, Jupiter and Saturn entry result in very high entry speeds, deceleration, and heating making them the most demanding destinations for atmospheric entry and impractical for aerocapture. Titan is a unique destination, with its low gravity and greatly extended thick atmosphere enabling low deceleration and heating loads for entry and aerocapture. Uranus and Neptune also have large gravity wells, resulting in high entry speeds, high deceleration and heating compared to the inner planets, but are still less demanding than Jupiter or Saturn., Comment: 15 pages, 18 figures

The ability to launch small secondary payloads to Mars on future science missions present an exciting opportunity for demonstration of advanced technologies for planetary exploration such as aerocapture. Over the years, several aerocapture technology demonstration missions have been proposed but none could be realized, causing the technology to become dormant as it is seen as too risky and expensive to be used on a science mission. The present study proposes a low-cost Mars aerocapture demonstration concept as a secondary payload, and could pave the way for future low-cost small interplanetary orbiter missions. The proposed mission heavily leverages the mission architecture and the flight hardware of the MarCO spacecraft for a low cost mission. The 35 kg technology demonstrator would launch as an ESPA secondary payload on a future Mars mission, and would be deployed from the upper stage soon after primary spacecraft separation. The vehicle then independently cruises to Mars, where it will perform aerocapture and insert a 6U MarCO heritage CubeSat to a 200 x 2000 km orbit. The mission architecture incorporates a number of cost saving approaches, and is estimated to fit within a 30M cost cap, of which 10M is allocated for technology development and risk reduction., Comment: 14 pages, 9 figures

Aerocapture is a technique which uses atmospheric drag to decelerate a spacecraft and achieve nearly fuel-free orbit insertion from an interplanetary trajectory. The present study performs a historical review of the field, and a bibliometric data analysis of the literature from 1980 to 2023. The data offers insights into the evolution of the field, current state of research, and pathways for its continued development. The data reveal a pattern in the rise of publications, followed by a period of stagnation, which repeats itself approximately once every decade. Mars is the most studied destination, while Uranus is the least studied. Prior to 2013, NASA centers produced the most publications and are the most cited in the field. However, academic institutions produced the majority of publications in the last decade. The United States continues to be the leading country in terms of publications, followed by China. The Journal of Spacecraft and Rockets is the leading source of publications, both in terms of number and citations. NASA is the leading funding source, followed by the National Natural Science Foundation of China. A proposed low-cost Earth flight demonstration of aerocapture will greatly reduce the risk for future science missions., Comment: 19 pages, 20 figures

The Cassini-Huygens mission has transformed our understanding of Titan from a hazy veiled moon to a place surprisingly like the Earth, with terrestrial physical processes such as wind, rainfall, and erosion shaping the landscape albeit with entirely different chemistry and temperatures. Dragonfly, a single element mission which fits within the New Frontiers cost cap will arrive at Titan in 2034, and perform in-situ investigations of the organic materials on the surface. However, its detailed investigations will be limited to region within its short flight range. The big gaps in our understanding of Titan global topography, climate, and upper atmospheric chemistry which can only investigated from an orbiter around Titan will remain to be addressed by a future orbiter mission. Due to the challenges of attaining orbit, past Titan orbiter concepts have been beyond the New Frontiers cost cap. The present study explores the use of drag modulation aerocapture for a Titan Orbiter which fits within New Frontiers. The study shows how a Dragonfly-like lander, and a Titan orbiter which each individually fit within the New Frontiers cost cap, when combined together can provide the science data return equivalent to a Flagship-class mission., Comment: 17 pages, 10 figures

Sleep plays a crucial role in the well-being of human lives. Traditional sleep studies using Polysomnography are associated with discomfort and often lower sleep quality caused by the acquisition setup. Previous works have focused on developing less obtrusive methods to conduct high-quality sleep studies, and ear-EEG is among popular alternatives. However, the performance of sleep staging based on ear-EEG is still inferior to scalp-EEG based sleep staging. In order to address the performance gap between scalp-EEG and ear-EEG based sleep staging, we propose a cross-modal knowledge distillation strategy, which is a domain adaptation approach. Our experiments and analysis validate the effectiveness of the proposed approach with existing architectures, where it enhances the accuracy of the ear-EEG based sleep staging by 3.46% and Cohen's kappa coefficient by a margin of 0.038., Comment: Code available at : https://github.com/Mithunjha/EarEEG_KnowledgeDistillation

Accurate sleep stage classification is significant for sleep health assessment. In recent years, several machine-learning based sleep staging algorithms have been developed, and in particular, deep-learning based algorithms have achieved performance on par with human annotation. Despite the improved performance, a limitation of most deep-learning based algorithms is their black-box behavior, which has limited their use in clinical settings. Here, we propose a cross-modal transformer, which is a transformer-based method for sleep stage classification. The proposed cross-modal transformer consists of a novel cross-modal transformer encoder architecture along with a multi-scale one-dimensional convolutional neural network for automatic representation learning. Our method outperforms the state-of-the-art methods and eliminates the black-box behavior of deep-learning models by utilizing the interpretability aspect of the attention modules. Furthermore, our method provides considerable reductions in the number of parameters and training time compared to the state-of-the-art methods. Our code is available at https://github.com/Jathurshan0330/Cross-Modal-Transformer., Comment: 11 pages, 7 figures, 6 tables

We revisit the mechanism of vortex unpinning caused by the neutron-vortex scattering \cite{prad1} in the inner crust of a pulsar. The strain energy released by the crustquake is assumed to be absorbed in some part of the inner crust and causes pair-breaking quasi-neutron excitations from the existing free neutron superfluid in the bulk of the inner crust. The scattering of these quasi-neutrons with the vortex core normal neutrons unpins a large number of vortices from the thermally affected regions and results in pulsar glitches. We consider the geometry of a cylindrical shell of the affected pinning region to study the implications of the vortex unpinning in the context of pulsar glitches. We find that a pulsar can release about $\sim 10^{11} - 10^{13}$ vortices by this mechanism. These numbers are equivalent to the glitch size of orders $\sim 10^{-11} - 10^{-9}$ for Vela-like pulsars with the characteristic age $\tau \simeq 10^4$ years. We also suggest a possibility of a vortex avalanche triggered by the movement of the unpinned vortices. A rough estimate of the glitch size caused by an avalanche shows an encouraging result., Comment: 12 pages, 5 figures, Accepted for publication in PRD

Human behavioral monitoring during sleep is essential for various medical applications. Majority of the contactless human pose estimation algorithms are based on RGB modality, causing ineffectiveness in in-bed pose estimation due to occlusions by blankets and varying illumination conditions. Long-wavelength infrared (LWIR) modality based pose estimation algorithms overcome the aforementioned challenges; however, ground truth pose generations by a human annotator under such conditions are not feasible. A feasible solution to address this issue is to transfer the knowledge learned from images with pose labels and no occlusions, and adapt it towards real world conditions (occlusions due to blankets). In this paper, we propose a novel learning strategy comprises of two-fold data augmentation to reduce the cross-domain discrepancy and knowledge distillation to learn the distribution of unlabeled images in real world conditions. Our experiments and analysis show the effectiveness of our approach over multiple standard human pose estimation baselines., Comment: Code is available at https://github.com/MohamedAfham/CD_HPE

Pulsars undergoing crustquake release strain energy, which can be absorbed in a small region inside the inner crust of the star and excite the free superfluid neutrons therein. The scattering of these neutrons with the surrounding pinned vortices may unpin a large number of vortices and effectively reduce the pinning force on vortex lines. Such unpinning by neutron scattering can produce glitches for Crab like pulsars and Vela pulsar of size in the range $\sim 10^{-8} - 10^{-7}$, and $\sim 10^{-9} - 10^{-8}$, respectively. Although we discuss here the crustquake initiated excitation, the proposal is very generic and equally applicable for any other sources, which can excite the free superfluid neutrons, or can be responsible for superfluid - normal phase transition of neutron superfluid in the inner crust of a pulsar., Comment: 9 pages, 1 figure, accepted for publication in Monthly notices of Royal Astronomical Society (MNRAS)

We explore here a possible consequence of crustquake, namely, the generation of bursts of gravitational waves (GWs) due to a sudden change in the quadrupole moment (QM) of a deformed pulsar as a result of crustquake. The occurrence of crustquake in a rotating neutron star can play many important roles in neutron star (NS) dynamics. Here we propose that if a pulsar undergoes crustquake, then the generation of bursts of GWs is an inevitable consequence of crustquake. We have estimated the strain amplitudes ($h_0$) for such bursts of GWs and compared with the strain amplitudes for GWs produced in various other scenarios for isolated pulsars as suggested earlier in the literature. The values we obtain are comparable to those suggestions. We also estimate the order of magnitude for characteristic strain ($h_c$) and signal to noise ratio (SNR) for such bursts. For exotic quarks stars, a multifold enhancement of strain amplitudes is expected, which makes quark stars a potential source of gravitational waves as a result of crustquake. The absence of such waves may put constraints on such hypothetical stars, Comment: 8 pages, 2 figures, 1 table. Accepted for publication in Journal of Astrophysics and Astronomy. Comments are welcome!