How Hive Scientific Separates the Good Science from the Bad
Over 1 million research papers are published each year. 60%-90% are not reproducible. Ahmed Alkhateeb discusses a potential solution.
By Elizabeth Wu
Ahmed Alkhateeb, PhD, Founder of Hive Scientific, sits down with Scismic to share his views on scientific automation and irreproducibility in biomedicine.
With over 25 million papers indexed in PubMed — and the number of publications increasing by more than 1 million each year — it is becoming more and more difficult for scientists to sort through the vast body of literature as they formulate their hypotheses.
Some reports find that 60%-90% of papers cannot be independently replicated. Valuable resources are wasted on unreliable experiments, which eventually contribute to the failure of clinical trials that attempt to build off these irreproducible findings.
With so much noise and non-reproducible data, how can scientists evaluate the quality and reproducibility of research performed?
Ahmed Alkhateeb is building a solution. In 2015, he founded Hive Scientific to identify reliable results from the 25 million papers published in biomedicine. It uses an analytical approach in order to generate objective quality metrics that rank scientific observations. “The idea was to create a better way to evaluate the science that goes beyond citations, impact factor, and reputation,” Alkhateeb says. “I wanted to increase the signal-to-noise ratio so we, as scientists, could come up with better hypotheses and become more efficient in researching biology.”
How Hive Scientific Works
Alkhateeb’s first challenge in building Hive was figuring out how to efficiently evaluate the scientific literature, and separate robust science from speculation. Through an introspective process, he was able to reduce the evaluation process into small steps that could be easily carried out. He then fragmented papers into their most basic components and integrated those components into an automated analytical engine. “This approach allows for the robust science to rise among the totality of observations within a field — rather than the status quo of highlighting the small fraction of papers that scientists happen to read. In a way, we are adding a layer of inductive logic to the whole thing,” Alkhateeb says.
Hive aims to serve the pharmaceutical industry by validating the robustness of the science behind potential drug targets. Only 10% of all drugs that go to clinical trials receive FDA approval. The Hive team has demonstrated that their algorithm can differentiate between a drug that is likely to succeed and a drug that is likely to fail in clinical trials based on the robustness of the findings behind each target.
In the future, Alkhateeb envisions Hive as an engine that automates scientific discovery. “The Hive database will not only include high-resolution scientific observations, but it would also be able to evaluate the strength of each observation. This approach overcomes a major hurdle in scientific AI — this is not ‘garbage in, garbage out.’ We would be able to generate highly plausible hypotheses. Scientists can then design experiments to test these hypotheses and feed them into robotic platforms to perform the experiments.”
“The scientific method needs an update,” Alkhateeb says. “If we can centralize the collection of scientific observations and automate hypothesis generation, we might be able to increase the rate of scientific discovery 10-fold.”
When asked how scientists can help create a world with more reliable science, Alkhateeb responded, “Young scientists who recognize that there’s a problem have to become more active. They have to turn into activist-scientists. It’s not about protesting, but being aware that there are people working on solutions that require their help.”
Alkhateeb urges scientists to become active and engaged with new initiatives proposed by entrepreneurs, which could help researchers in the short-term, and will improve science in the long-term. He asks researchers to take action instead of accepting that the system is unfair.
“The problems that face science are a product of a dysfunctional system that can be fixed. They are not laws of nature.”
Scismic fosters groundbreaking science by connecting researchers to jobs that empower them to perform their best. Scismic Lab Seeker, our free academic lab database, will help you find the perfect academic lab! Sign up today!