Small molecule autoencoders: architecture engineering to optimize latent space utility and sustainability

Autoencoders are frequently used to embed molecules for training of downstream deep learning models. However, evaluation of the chemical information quality in the latent spaces is lacking and the model architectures are often arbitrarily chosen. Unoptimized architectures may not only negatively affect latent space quality but also increase energy consumption during training, making the models unsustainable. We conducted systematic experiments to better understand how the autoencoder architecture affects the reconstruction and latent space quality and how it can be optimized towards the encoding task as well as energy consumption. We can show that optimizing the architecture allows us to maintain the quality of a generic architecture but using 97% less data and reducing energy consumption by around 36%. We additionally observed that representing the molecules as SELFIES reduced the reconstruction performance compared to SMILES and that training with enumerated SMILES drastically improved latent space quality. Scientific Contribution: This work provides the first comprehensive systematic analysis of how choosing the autoencoder architecture affects the reconstruction performance of small molecules, the chemical information content of the latent space as well as the energy required for training. Demonstrated on the MOSES benchmarking dataset it provides first valuable insights into how autoencoders for the embedding of small molecules can be designed to optimize their utility and simultaneously become more sustainable, both in terms of energy consumption as well as the required amount of training data. All code, data and model checkpoints are made available on Zenodo (Oestreich et al. Small molecule autoencoders: architecture engineering to optimize latent space utility and sustainability. Zenodo, 2024). Furthermore, the top models can be found on GitHub with scripts to encode custom molecules: https://github.com/MarieOestreich/small-molecule-autoencoders .

An efficient algorithm for resource optimization in IRS-mmWave-NOMA B5G wireless networks

The effectiveness of implementing intelligent reflecting surface (IRS) for millimeter-wave (mmWave)-non-orthogonal multiple-access (NOMA) systems has allowed for significant sum-rate improvements. The majority of recent research has not discussed how well the IRS-mmWave-NOMA combination performs. Therefore, a new technique for resource optimization in IRS-mmWave-NOMA B5G wireless networks is proposed in this research. The key concept is to use an iterative algorithm to solve the optimization issue while incorporating many crucial constraints like the selection of the IRS beam, transmit power distribution, and decoding order, among others. Simulation results show that the proposed approach outperforms existing state-of-the-art algorithms in terms of computation delay, sum rate and NMSE. The computational complexity also validated the simplicity and hardware-friendly feature of the proposed algorithm.

An efficient algorithm for energy harvesting in IIoT based on machine learning and swarm intelligence

The Internet of Things (IoT) is a network of smart gadgets that are connected through the Internet, including computers, cameras, smart sensors, and mobile phones. Recent developments in the industrial IoT (IIoT) have enabled a wide range of applications, from small businesses to smart cities, which have become indispensable to many facets of human existence. In a system with a few devices, the short lifespan of conventional batteries, which raises maintenance costs, necessitates more replacements and has a negative environmental impact, does not present a problem. However, in networks with millions or even billions of devices, it poses a serious problem. The rapid expansion of the IoT paradigm is threatened by these battery restrictions, thus academics and businesses are now interested in prolonging the lifespan of IoT devices while retaining optimal performance. Resource management is an important aspect of IIoT because it's scarce and limited. Therefore, this paper proposed an efficient algorithm based on federated learning. Firstly, the optimization problem is decomposed into various sub-problems. Then, the particle swarm optimization algorithm is deployed to solve the energy budget. Finally, a communication resource is optimized by an iterative matching algorithm. Simulation results show that the proposed algorithm has better performance as compared with existing algorithms.

5G standards for the Industry 4.0 enabled communication systems using artificial intelligence: perspective of smart healthcare system

The emergence of the Industry 4.0 revolution to upgrade the Internet of Things (IoT) standards provides the prominence outcomes for the future wireless communication systems called 5G. The development of 5G green communication systems suffers from the various challenges to fulfill the requirement of higher user capacity, network speed, minimum cost, and reduced resource consumption. The use of 5G standards for Industry 4.0 applications will increase data rate performance and connected device's reliability. Since the arrival of novel Covid-19 disease, there is a higher demand for smart healthcare systems worldwide. However, designing the 5G communication systems has the research challenges like optimum resource utilization, mobility management, cost-efficiency, interference management, spectral efficiency, etc. The rapid development of Artificial Intelligence (AI) across the different formats brings performance enhancement compared to conventional techniques. Therefore, introducing the AI into 5G standards will optimize the performances further considering the various end-user applications. We first present the survey of the terms like 5G standard, Industry 4.0, and some recent works for future wireless communications. The purpose is to explore the current research problems using the 5G technology. We further propose the novel architecture for smart healthcare systems using the 5G and Industry 4.0 standards. We design and implement that proposed model using the Network Simulator (NS2) to investigate the current 5G methods. The simulation results show that current 5G methods for resource management and interference management suffer from the challenges like performance trade-offs.

Analyses on ICU and non-ICU capacity of government hospitals during the COVID-19 outbreak via multi-objective linear programming: An evidence from Istanbul

The current infectious disease outbreak, a novel acute respiratory syndrome [SARS]-CoV-2, is one of the greatest public health concerns that the humanity has been struggling since the end of 2019. Although, dedicating the majority of hospital-based resources is an effective method to deal with the upsurge in the number of infected individuals, its drastic impact on routine healthcare services cannot be underestimated. In this study, the proposed multi-objective, multi-period linear programming model optimizes the distribution decision of infected patients and the evacuation rate of non-infected patients simultaneously. Moreover, the presented model determines the number of new COVID-19 intensive care units, which are established by using existing hospital-based resources. Three objectives are considered: (1) minimization of total distance travelled by infected patients, (2) minimization of the maximum evacuation rate of non-infected patients and (3) minimization of the infectious risk of healthcare professionals. A case study is performed for the European side of Istanbul, Turkey. The effect of the uncertain length of the stay of infected patients is demonstrated via sensitivity analyses.

Impact of the subcutaneous formulations of trastuzumab and rituximab on efficiency and resource optimization in Spanish hospitals: H-Excelencia study

Background

Subcutaneous (SC) versus intravenous (IV) administration is advantageous in terms of patient convenience and hospital efficiency. This study aimed to compare the effect of optimizing the processes involved in SC versus IV administration of rituximab and trastuzumab on hospital capacity and service quality.

Methods

This cross-sectional resource utilization study interviewed oncologists, hematologists, nurses, and pharmacists from 10 hospitals in Spain to estimate changes in processes associated with conversion from IV to SC rituximab and trastuzumab, based on clinical experience and healthcare use from administrative databases.

Results

Efficient use of SC formulations increased the monthly capacity for parenteral administration by 3.35% (potentially increasable by 5.75% with maximum possible conversion according to the product label). The weekly capacity for hospital pharmacy treatment preparation increased by 7.13% due to conversion to SC formulation and by 9.33% due to transferring SC preparation to the cancer treatment unit (potentially increasable by 12.16 and 14.10%, respectively). Monthly hospital time decreased by 33% with trastuzumab and 47% with rituximab. In a hypothetical hospital, in which all processes for efficient use of SC rituximab and/or trastuzumab were implemented and all eligible patients received SC formulations, the estimated monthly capacity for preparation and administration increased by 23.1% and estimated hospital times were reduced by 60–66%.

Conclusions

Conversion of trastuzumab and rituximab to SC administration could improve the efficiency of hospitals and optimize internal resource management processes, potentially increasing care capacity and improving the quality of care by reducing time spent by patients at hospitals.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12913-021-06277-8.

Designing a novel hybrid healthcare teleconsultation network: a benchtop study of telepathology in Iran and a systematic review

BACKGROUND

Growing demand for medical services has increased patient waiting time due to the limited number or unbalanced distribution of healthcare centers. Healthcare teleconsultation networks are one of the potentially powerful systems to overcome this problem. Medical pathology can hugely benefit from teleconsultation networks because having second opinions is precious for many cases; however, resource planning (i.e., assignment and distribution of pathology consultation requests) is challenging due to bulky medical images of patients. This results in high setup and operational costs. The aim of this study is to design an optimal teleconsultation network for pathology labs under the supervision of medical sciences universities in Tehran, Iran.

METHODS

To avoid the setup cost, we first propose a modified hybrid peer-to-peer (P2P) overlay architecture for our telepathology network, using Iran's National Healthcare Information Network (SHAMS) as the underlying infrastructure. Then we apply optimization techniques to solve the request assignment and distribution problems in the network. Finally, we present a novel mathematical model with the objective of minimizing the variable operational costs of the system.

RESULTS

The efficiency of the proposed method was evaluated by a set of practical-sized network instances simulated based on the characteristics of SHAMS. The results show that the presented model and architecture can obtain optimal solutions for network instances up to 350 nodes, which covers our target network.

CONCLUSIONS

We believe that the proposed method can be beneficial for designing large-scale medical teleconsultation networks by adjusting the constraints according to the rules and conditions of each country. Our findings showed that teleconsultation networks in countries with strong information technology (IT) infrastructures are under the influence of consultation fees, while in countries with weak IT infrastructure, the transmission costs are more critical. To the best of our knowledge, no research has so far addressed resource planning in medical teleconsultation networks using optimization techniques. Besides, the target network, i.e., pathology labs under the supervision of medical sciences universities in Tehran and the SHAMS network, are discussed for the first time in this work.

[Emergency room activation due to trauma mechanism]

BACKGROUND

The quality of trauma care in Germany has been significantly increased due to the establishment of standards in the white paper on severe injury care and the S3 guidelines. A key issue of multiple trauma treatment is the trauma resuscitation unit (TRU)/emergency room management, which is associated with extensive material and human resources. From the very beginning of the introduction of structured care for the severely injured, the choice of the target hospital and the indications for TRU have been the focus of scientific research. Furthermore, a reduction of the TRU team for presumably less seriously injured patients is discussed.

MATERIAL AND METHODS

The emergency room assignments of a level I trauma center (n = 686) were analyzed in more detail. Of the patients 235 were assigned with the TRU indications according to the cause of the accident (GoR B criteria) and compared with the collective of TRU patients admitted according to the severity of injuries or life-threatening signs, the so-called GoR A criteria (n = 104) during the corresponding period. In addition to basic data (age, sex), the injured region and severity (injury severity score, ISS), the length of stay in the intensive care unit (ICU) and hospital as well as the necessity for surgery and transfusion were compared.

RESULT

Of the emergency room allocations at the trauma center 34% were due to the cause of the accident and the severity of the injuries in this patient group was almost half as high as that of the control group with an ISS of 11. Of the patients 74% were admitted to the IMC/ICU and stayed there for an average of almost 3 days. There were between 4% and 18% severe injuries (abbreviated injury scale, AIS 3) and 17.9% were characterized as polytrauma with an ISS ≥ 16 points.

CONCLUSION

A significant number of patients admitted to a TCU due to the cause of accident (the so-called B criteria of the white book), have severe and potentially life-threatening injuries, which necessitate a prioritized and immediate treatment by a TCU team. Whether a reduced TCU team is sufficient in this situation needs to be critically examined.

Multidisciplinary approach to psoriasis in the Spanish National Health System: A social return on investment study

Introduction

Psoriasis is a chronic disease in which patients feel stigmatization, social rejection, and suffer from low self-esteem. There are still unmet needs that make it necessary to define a new multidisciplinary approach to provide benefits not only to patients and their families but also to the Spanish National Health System (SNHS) and society. The aim was to define a new approach to better address the unmet needs of patients with psoriasis within the SNHS and to measure its impact from a social perspective, that is, in clinical, health care, economic, and social terms.

Methods

Multidisciplinary experts identified, agreed on, and selected several health care interventions that were feasible for implementation in the SNHS. This process was carried out in four different areas: diagnosis, mild psoriasis, moderate psoriasis, and severe psoriasis. To estimate investment and social return, the social return on investment (SROI) method was used.

Results

The new approach to psoriasis management in the SNHS comprised 18 proposals. The investment needed for the implementation of this new approach would amount to €222.77 million and its return to €1,123.11 million. This would yield a SROI ratio of €5.04 for every euro invested.

Conclusion

The new approach to psoriasis management would yield a positive social return. The results will allow optimal strategic planning adapted to each assistance situation, to achieve a comprehensive and multidisciplinary approach.

A water-filled garment to protect astronauts during interplanetary missions tested on board the ISS

As manned spaceflights beyond low Earth orbit are in the agenda of Space Agencies, the concerns related to space radiation exposure of the crew are still without conclusive solutions. The risk of long-term detrimental health effects needs to be kept below acceptable limits, and emergency countermeasures must be planned to avoid the short-term consequences of exposure to high particle fluxes during hardly predictable solar events. Space habitat shielding cannot be the ultimate solution: the increasing complexity of future missions will require astronauts to protect themselves in low-shielded areas, e.g. during emergency operations. Personal radiation shielding is promising, particularly if using available resources for multi-functional shielding devices. In this work we report on all steps from the conception, design, manufacturing, to the final test on board the International Space Station (ISS) of the first prototype of a water-filled garment for emergency radiation shielding against solar particle events. The garment has a good shielding potential and comfort level. On-board water is used for filling and then recycled without waste. The successful outcome of this experiment represents an important breakthrough in space radiation shielding, opening to the development of similarly conceived devices and their use in interplanetary missions as the one to Mars.