Boosting the Science: Get Rid of the Wires

Back to the future

“The greater the spaces to be bridged, the more rational becomes communication without wires.”

— Nikola Tesla, The Problem of Increasing Human Energy, 1900 [1]

So wrote Nikola Tesla in his revolutionary article in 1900. Among other futuristic visions, he presented a “world system” based on wireless communication — what we would now understand as the internet. Tesla predicted that it would allow us to send any amount of almost any kind of information to any distance. “When wireless is fully applied, the earth will be converted into a huge brain, capable of response in every one of its parts,” he wrote to J.P. Morgan, one of the world’s most powerful men at that time.

Now, over a hundred years later, we can see that this great historical vision is either idealistic or still unfulfilled. There are still many wires. However, we should not understand a “wire” literally as a flexible thread or slender rod of metal [3]. In the 21st century, a metaphorical “wire” is every hindrance to development. Many of them are not as evident as the wires in the wall.

Meanwhile, the accelerating development of diverse technologies, a growing human population and the increasing speed of life contribute to overlooking multiple inaccuracies. Unnoticed, they accumulate and cause large-scale, global problems, as they are continuously generated by fundamental, yet already old-fashioned solutions. Innovation is not just creating something new. The times of great revolutions have passed and led us to the days of necessary precision. Innovation means using existing solutions to release development from its hindrances. These problems do not concern the newest technologies, the biggest companies or the greatest countries. The real reasons for global problems are hidden deep in our habits, our simplest imperfections and the limitations of our brains. Technology can help us find them and improve.

The vision of our planet as a huge brain with boundless connectivity is realistic, yet difficult to accomplish. We already live in a world of Big Data. The expanding data collection generates vast spaces that need to be bridged. This is possible with our technology, with sophisticated algorithms and innovative solutions. However, Big Data include False Data, and connecting the two might bring terrifying consequences — some of which we already experience [4].

Distinguishing the existing false from the true is our great challenge. But as we all know, prevention is better than cure. False data can be eliminated before they appear, or recognized right after they occur and pushed aside. Both solutions are widespread in Nature, which is the best teacher inspiring us to create solutions.

Consider the human brain, which comprises approximately 86 billion neurons that form a web of about 100 trillion synapses [5]. Each synapse is highly specific. Its formation is preceded by the approach of two particular neurons that meticulously search for the unique location of the best connection between one neuron’s axon terminal and another neuron’s dendritic spine. The result is a highly specialized synapse, characterized by a specific combination of expressed receptors and neurotransmitters — the information molecules released by one neuron and received by the other. Even though many molecules and mechanisms are involved in forming this complex structure, some synapses are naturally destroyed so that we can learn effectively, combine our knowledge and forget what is unimportant. These phenomena happen all the time in our brains. They show how dynamic change in search of improvement can bring amazing results. There are no stable states in nature.

“But articles are stable. Paper is stable. These are the exemplary wires that slow down our progress.”

Getting rid of just these could accelerate our progress, increase our life satisfaction, eliminate endless paper documentation and queues in institutions, improve health care management and, finally, notably improve the quality of scientific research. Information is energy which, when trapped in a vastness of other information, can easily be omitted and forgotten, like a sentence in a book. Whenever it is released to flow freely, it can be used every time and everywhere it is needed, and hence actively contribute to a flexible entirety of human knowledge. Presumably, this is the Earth as “a huge brain” that Tesla envisioned.

“Though free to think and act, we are held together, like the stars in the firmament, with ties inseparable. These ties cannot be seen, but we can feel them. (...) And now I see stricken down an enemy, a lump of matter which, of all the lumps of matter in the universe, I care least for, and it still grieves me. Does this not prove that each of us is only part of a whole?”

— Nikola Tesla [1]

Wires in the science

We face a global crisis of biomedical research reproducibility. It is estimated that about 70% of scientific reports, including those in journals with a high impact factor, are irreproducible [6–11]. An economic analysis carried out in 2015 indicated that in the USA, 28 billion dollars a year are spent on preclinical research that cannot be reproduced, assuming a 50% irreproducibility rate [12].

Presumably, the biggest problem leading to the reproducibility crisis is data loss. Selective reporting is common in biomedical science [13–17]. It includes:

• Lack of standards in data recording
• Imprecise planning of experiments
• Incomplete reporting of results and methodology
• Excluding results that do not fit the research hypothesis

Another problem affecting reproducibility is the pressure to publish [18,19]. For the last 250 years, the number of scientific publications has been rising exponentially, approaching two million papers per year by 2012 [18]. Moreover, publishing positive results is a truth-distorting trend which increased by 22% between 1990 and 2007 [20].

In summary, a high publication frequency — especially of positive results, preferentially published in high-IF journals and resulting in a high volume of citations — gives a scientist better chances of receiving funds for even more intriguing and innovative projects. The pressure is high.

Meanwhile, the quality of published preclinical data contributes to the scale of failure in clinical trials. A report from 2011 presents dramatically low and, what is more worrying, decreasing success rates of Phase II clinical trials [22].

The statistics problem

The next big problem is statistical data analysis. Many scientists use their own knowledge to handle the analysis before (though this does not always happen) and after the experiment. However, statistics is a science in itself and should be carried out by professionals. A great issue is the widespread misuse and abuse of statistical significance by scientists.

To show how misleading the p-value can be, Joseph Simmons, an experimental psychologist from the University of Pennsylvania, carried out an experiment revealing how simple it is to find a statistically significant result that “proves” any hypothesis. He demonstrated that the age of the experiment’s participants decreased by 1.5 years after listening to the song “When I’m Sixty-Four” by The Beatles [23].

A historic publication, “Why Most Published Research Findings Are False” by Professor John P.A. Ioannidis, reached the highest popularity in the history of the Public Library of Science (PLOS), exceeding 1.5 million views after its publication in 2005. Professor Ioannidis uses statistics to demonstrate the low probability that scientific reports are reliable. He points to six problems that contribute to this worrying phenomenon [24]:

• Small sample sizes
• Small observed effects
• A large number and small pre-selection of tested relationships
• High flexibility in study design, definitions, outcomes and analysis methods
• Conflicts of interest (especially financial) and scientific bias
• The size of scientific consortium working on the same issue (multi-group research)

Science as a collective project

Humans undertake big, challenging projects one after another — and science should constantly be regarded as one of them. Making science a collective project could eliminate many of the problems mentioned above, like small sample sizes. Altogether, there are millions of scientists around the globe who are often unaware of others working on the same problem. A global science management system could prevent wasted time, materials and funds. It could connect scientists, verify their hypotheses and combine results.

If every result of scientific research were presented as part of a global mathematical model of the particular phenomenon, we would avoid isolated individual analyses by scientists. Models would need to be built by authorized professionals responsible for maintaining the highest quality. Releasing scientists from obligatory written reports would eliminate fraud, competition and the rat race for the largest number of publications. Scientists could spend more time on actual research, and their results could be combined into a global scientific knowledge database of interdependent information.

Software

“When we speak of man, we have a conception of humanity as a whole, and before applying scientific methods to the investigation of his movement, we must accept this as a physical fact.”

— Nikola Tesla [1]

Scientific research is not work done for a single company but for the whole of humanity. Therefore, no scientific effort should be confidential. Fortunately, more and more people want to turn the science of secrets into open access science. We stand at the front of a scientific revolution. Making raw data, precise methodology, information on the materials used and analytic algorithms fully accessible gives endless possibilities:

• Data comparison and analysis across studies
• Creating global standards
• Facilitating learning
• Better hypothesis formation and verification
• Finding and fixing mistakes — and, in consequence, higher-quality research

This revolution is inevitable, and it is finally getting started. In April 2016, the European Commission presented its blueprint for a new European Open Science Cloud (EOSC), a virtual environment where millions of scientists can store, share, access and process large datasets with supercomputer capacity. In May 2017, the Commission published the book Open Innovation, Open Science, Open to the World, highlighting key conceptual insights into Europe’s future of research and innovation [25]. Big scientific databases are already available online, and there are algorithms that search for specific information among thousands of articles to run a statistical analysis of a phenomenon, or to personalize your search and suggest articles you might be interested in.

Beyond the written word

However, collecting Big Data, processing it and extracting astonishing conclusions or predictions is just the beginning. The revolution goes far beyond this. Imagine replacing the written word, to a great extent, with images or process simulations. The written word is the past. The fastest ideas appearing in our minds do not show up as words; they are more related to images and feelings. Therefore, structural, molecular and finally functional models of different phenomena are of great interest.

A noteworthy example is the work of the Polish scientist Professor Wiesław Nowiński, who developed 35 three-dimensional brain models. These structural brain atlases combine multiple MRI scans of his own brain. They are used to explore the brain’s vascular system, as well as for the diagnosis, prediction and treatment of various brain disorders, including stroke, Parkinson’s and Alzheimer’s diseases, and psychiatric conditions. Patients’ scans are superimposed on the brain model to compare corresponding tissue structures and find deviations from healthy tissue.

It took many years to create these beautiful and highly useful atlases. But such models could be built continuously. Every MRI performed anywhere in the world could be collected by a global system as anonymous information of scientific value, described by the patient’s basic parameters and diagnosis code. By comparing these scans, we could create a vast model of any tissue and its differences depending on sex, age and genetic origin. We could trace healthy and disordered tissue development, and precisely predict diseases or susceptibility to specific, rare conditions. We could turn rare conditions into well-known ones. This can be done with any part of our bodies. We could combine different diagnostic methods to generate complete models and create a priceless source of accumulated medical knowledge. Adding artificial intelligence could produce a virtual doctor, constantly updated with the most current knowledge and accessible to everyone.

After the Human Genome Project, which described the structure of our genome, we face another great challenge: the Human Cell Atlas Project. Using single-cell molecular profiling, scientists aim to define all human cell types in terms of molecular profiles, such as gene expression patterns, combined with the morphology and location of the cells. The project is very ambitious, especially given the compelling concept of somatic mosaicism, which indicates that no two cells in an organism are the same [27,28].

From structure to function: a virtual cell

However, structural models are not enough. To really understand our nature, we need functional models that show processes, especially at the molecular level. This involves molecular modeling. The preclinical results available in articles could be transformed into “living” maps of molecular relationships. We already have big maps of molecular mechanisms, but this can go much further. Compressing all existing information about one phenomenon would give its multidimensional model. Connecting known phenomena by their relationships would generate a virtual world of the cell’s interior.

We could simulate the processes happening in the cell and, for example, see what our in vitro study looks like from inside the cell. In every detail, we could see which protein was activated, what its structure is, where it is located, how it changes its shape, how it affects the shapes of surrounding proteins and what the dynamics of these actions are. In this virtual environment we could decide which information to explore in depth. We could see the general mechanism and watch the cell’s inner life like a movie — or stop at one point and explore every available piece of data about a particular element, including the sources used to build that part of the model. The information could be written in articles, shown in additional images or models, or presented as a movie or documentary — like references in articles. Further, we could run basic experiments virtually, using worldwide knowledge accumulated and processed over many years.

This virtual world can be generated using virtual and augmented reality technologies, which are already used in many aspects of our lives. In education, medical students can learn the structures of diverse organs without entering the dissection room, and train surgeries on virtual patients, using real tools while watching everything on a screen. These technologies enable remote surgeries that could serve soldiers on a mission or astronauts on space flights. Combined with haptic technology, they allow a surgeon to feel what they are working on. In the laboratory, we could use these realities to simulate experiments — for example, to learn how a mutation affects cell morphology, metabolism or life cycle. Over time, the available information will become more and more comprehensive, eventually generating a virtual, functional model of the human organism.

Hardware

The quality of scientific research depends on many factors, but one is inevitable: human error. We are a source of contamination, inaccuracy and mistakes. Compared to machines, we are slow and irreproducible. Diverse machines and robots already help us with laboratory work, but they are still just one part of the overall, mechanical laboratory activity. Nowadays, it takes many years of investigation to produce a cure for a disease. For some sick people, that is much too long. If we shortened this time two- or threefold, we could save many lives. It is possible.

The self-running laboratory

In the future, the laboratory should work by itself as a group of optimized, compatible technological solutions — automats and robots designed as a network of machines with complementary functionalities, managed by intelligent systems. A kind of high-quality, high-performance research factory that is:

• Capable of continuous operation according to human-defined protocols
• Isolated, to eliminate contamination and maintain the required conditions
• Remotely controlled
• Equipped with mathematical engines for complex data analysis
• Able to compare results with available knowledge, show tendencies and make predictions

Computing power for every scientist

A common problem of Big Data analysis is computing power. One solution is a supercomputer such as NASA’s Pleiades in Silicon Valley. Supercomputers deliver a high level of computing performance for visualizing extremely complex problems: tracing global climate and ocean patterns, weather modeling, physical simulations of aerodynamics to design safer aircraft and spacecraft, molecular modeling to help solve scientific problems and many others [29]. A supercomputer unique on a world scale was built in 2015 in the Polish BioNanoPark in Łódź. The Analyzer of Real Complex Systems (ARUZ) is the world’s largest simulator of physico-chemical processes, able to analyze more than one million molecules in a single test. This technology allows us to explore phenomena unseen in the laboratory [30].

However, such supercomputers are huge and extremely expensive, so they are not for common use. But technology is all around us. Nowadays, almost everyone has a mobile phone with many times more computing power than the computer on board Apollo 11. Together, these phones form part of a vast, potentially available distributed processing environment [31]. Any complex problem could be divided by a system and sent to a defined number of mobile phones, which would calculate it in seconds as background computation. The results would then be sent back to the system, which consolidates the divided data into a final result. This approach offers a tool for deep analysis to every scientist.

Combining robotic performance with fast, complex Big Data analysis could substantially increase quality and dramatically shorten the time needed to produce medical solutions for today’s untreatable disorders and diseases. We could save countless human lives.

Conclusions

“The scientific man does not aim at an immediate result. He does not expect that his advanced ideas will be readily taken up. His work is like that of the planter — for the future. His duty is to lay the foundation for those who are to come, and point the way.”

— Nikola Tesla [1]