How artificial intelligence is being used to speed trials
New insights from MoBa: An analysis of 10 genes that were altered during the birth of a baby by a woman in a healthy environment
Håberg conducted her postdoctoral work in the United States, where she joined a group at the National Institute of Environmental Health Sciences in Durham, North Carolina. She contributed data analysis to a team that examined 1,062 blood samples from MoBa, drawn from the umbilical cord at the delivery of a baby, and identified 10 genes that were altered in infants born to women who smoked while pregnant. The 2012 study provided important evidence that non-heritable smoking exposure can cause certain changes in the genome, but it does not change the sequence of the DNA. Hberg says they are only beginning to understand the gravity of epigenetic changes during development.
Hberg is passionate about connecting specialists from her team with groups from around the world so that they can explore large amounts of data that holds clues about fetal health. The MoBa data is being compared with information from the national birth cohort. She says it all comes down to finding exciting new ways to work together. It is good to see so much dedicated to early embryonic development. Amy Coombs is a person.
Source: Four change-makers seek impact in medical research
Artificial Intelligence in Computational Pathology: The Case of Analysing Cancerous Tumors and Mutations Using Deep Learning
As a university student studying medical photonics in Jena, Germany, Narmin Ghaffari Laleh was inspired to use her programming skills to help patients and doctors. She wanted to try out the use of artificial intelligence to improve medical-image analysis. Her work there concentrated on eye imaging, where conventional methods of analysis use systems that read each row of pixels, identifying features such as the cornea, lens and retina by tracking their colours and the distance between them. Common variables such as glasses can throw such systems off, however. “These kinds of programs work well until someone puts on glasses or contact lenses and takes a photo,” says Ghaffari Laleh, who was a master’s student at Fredrich Schiller University of Jena at the time.
Computational pathology is an emerging area of research that hopes to improve patient care by using new technologies such as artificial intelligence and big data. Her focus was on developing systems that can more accurately and efficiently identify visual indicators of cancer and other diseases than methods that rely solely on human specialists. These systems could be particularly useful in the analysis of tissue samples that have been prepared for microscope slides and stained with the widely used haematoxylin and eosin (H&E) dye, which turns cell structures different shades of purple, blue and pink, she says.
For a separate study6, the team showed how AI trained to identify mutations in a protein associated with bladder cancer could outperform a uropathologist in analysing tissue samples stained with H&E. “We don’t aim to replace the urologist but deep-learning can offer additional analysis.”
To test whether these methods can move to clinical applications, Ghaffari Laleh dedicated her PhD thesis to investigating how applicable these kinds of AI systems could be to a variety of diseases and patient demographics. She has a pending defence for her thesis in March.
Ghaffari Laleh hopes to apply her skills to help medical professionals in developing countries who cannot afford to run advanced diagnostics and who struggle to recruit and train skilled professionals. “AI is a much more affordable option,” she says. If a deep- learning model can analyse data from diverse patient groups from a wide range of countries, hospitals that don’t have the money can send samples to diagnose. She’s also working on AI that can read text7, ultrasound and radiology image data, with hopes that they can speed up the work of doctors and other specialists worldwide. — Amy Coombs
Having the capacity to launch research projects quickly proved invaluable to Patalon and her team during the COVID-19 pandemic, when global treatment and vaccination protocols changed rapidly to keep up with the evolution of the disease. The team of health professionals led by Pata Leon scoured the records of more than 100,000 Israelis to find out about the incidence and symptoms of coronaviruses in three months.
The team discovered that vaccinated people who had not previously tested positive for COVID-19 were 13 times more likely to be infected by the new variant, compared with previously infected individuals who were unvaccinated. The results showed that the natural immunity that will be provided to those who have been affected by the COVID-19 virus isn’t an immediate priority. It was big for us, and we were happy about it.
Extracting new insights from the vast amounts of public-health data that are being collected globally is key to advancing treatments and keeping one step ahead of infectious diseases, says Patalon. As part of her role at KSM, she oversees the Tipa Biobank, Israel’s largest biosample repository, comprising more than one million blood samples from some 200,000 Maccabi patients. In addition to one-off samples from patients, the biobank collects serial samples — successive samples from the same patient over a period of time. Serial samples are “very rare and highly valuable for research”, says Patalon, especially when it comes to analysing biological changes before and after a diagnosis.
Using the tuberal nucleus in the brain as a regulatory centre to induce mice to eat even if they weren’t hungry
Being a good leader and researcher is essential in times of crisis, as shown by the team’s reaction to the war in Gaza.
There is a lot of patience, empathy, emotional support and building of good relationships needed at this time. We have to come out of this situation stronger.” — Sandy Ong
Sacks brought a humanizing perspective to brain disorders and he was known for his empathizing approach to patients. “He showed how even minute changes in certain regions of the brain could lead to profound effects on cognition and behaviour.”
For many years researchers assumed that hunger is regulated by two types of cells, one driving hunger and the other suppressing it. When the researchers stimulated the tuberal nucleus in the brain, they were able to prompt mice to start eating even if they were not hungry. “There are actually many feeding regulatory centres in the brain, and we discovered one of them,” she says.
There are more diverse aspects of eating behavior that the other centres can deal with. In a series of follow-up experiments10, Luo and her colleagues observed that when mice were placed in the same feeding chamber where the neurons in the tuberal nucleus had been activated the previous week, they would immediately start eating, even if it was outside their normal feeding times. The results suggest that these neurons influence the basic feeding and eating processes.
He says that the environment you are in could cause your neurons to become activated. If those signals can cause you to eat, I think so.
“It would be very invasive to implant an electrode in the brain to activate or inhibit these pathways,” says Luo. But activating pathways that connect to these regions in the brain — by using vagal nerve stimulation, for example, which is a technique used to treat epilepsy that involves implanting a pulse generator under the skin on the chest — would be a more viable option. “Then maybe there will be an easier route for developing therapies to target some of these metabolic diseases,” says Luo. Sandy Ong.
Searching for Trials Using Artificial Intelligence and the Saama Appraisal System: Application to Predicting Patients’ Responses in a Clinical Trial
Predicting which patients will need a nudging can be done by other tools developed by Saama. Its tools can also combine all the data from a patient — such as lab tests, stats from wearable devices and notes — to assess outcomes. Moneymaker says it is not currently possible to analyse by hand the picture of an individual patient.
Helping people find each other doesn’t just speed up research. It makes it stronger. Artificial intelligence can find ways to include people who are pregnant, the elderly and children in trials. People with terminal cancer and those with rare diseases have an especially hard time finding trials to join. The patients sometimes do more work than the clinicians do. Match them with projects that are relevant to them.
Weng has also developed methods to help patients look for trials. One system, called DQueST, has two parts. The first uses Criteria2Query to extract criteria from trial descriptions. Patients can narrow their search with the help of questions generated by the second part. Another system, TrialGPT, from Sun’s lab in collaboration with the US National Institutes of Health, is a method for prompting a large language model to find appropriate trials for a patient. Given a description of a patient and clinical trial, it first decides whether the patient fits each criterion in a trial and offers an explanation. It then aggregates these assessments into a trial-level score. It does this for many trials and ranks them for the patient.
During a study or in a clinical practice, Chatbots can answer patients’ questions. One study took questions and answers from the AskDocs forum on www.redondo.com. A large percentage of the time, health-care professionals preferred the answers from the doctors. In another study5, researchers created a tool called ChatDoctor by fine-tuning a large language model (Meta’s LLaMA-7B) on patient-doctor dialogues and giving it real-time access to online sources. Questions about medical information that was newer than the training data could be answered by the ChatDoctor.
The most time consuming part of a clinical trial is recruiting patients, which takes up to a third of the study time. One in five trials don’t recruit the right number of people, and nearly all trials exceed expected recruitment timelines. Some researchers would like to accelerate the process by relaxing some of the eligibility criteria while maintaining safety. A group led by James Zou, a biomedical data scientist, has developed a system to analyze completed clinical trials and assesses how adjusting criteria for participation affects hazard ratios. In one case, they used it to run drug trials for lung cancer. They found that altering the criteria would have doubled the number of patients who were eligible, without increasing the hazard ratio. The study showed that the system also worked for other types of cancer and actually reduced harmful outcomes because it made sicker people — who had more to gain from the drugs — eligible for treatment.
A company called Intelligent Medical Objects in Rosemont, Illinois, has developed SEETrials, a method for prompting OpenAI’s large language model GPT-4 to extract safety and efficacy information from the abstracts of clinical trials. This lets trial designers see how other research has been used to design trials. The lab of Michael Snyder, a geneticist at Stanford University in California, developed a tool last year called CliniDigest that simultaneously summarizes dozens of records from ClinicalTrials.gov, the main US registry for medical trials, adding references to the unified summary. They’ve used it to summarize how clinical researchers are using wearables such as smartwatches, sleep trackers and glucose monitors to gather patient data. Alexander says that he has had conversations with many practitioners who see the potential in trials, but don’t know how to use them for highest impact. The field is moving so fast that it is not yet possible to find best practice.
The first step of the clinical-trials process is trial design. What amount of drugs should be given? To how many patients? What data should be collected on them? The lab of Jimeng Sun, a computer scientist at the University of Illinois Urbana-Champaign, developed an algorithm called HINT (hierarchical interaction network) that can predict whether a trial will succeed, based on the drug molecule, target disease and patient eligibility criteria. They followed up with a system called SPOT (sequential predictive modelling of clinical trial outcome) that additionally takes into account when the trials in its training data took place and weighs more recent trials more heavily. Based on the predicted outcome, pharmaceutical companies might decide to alter a trial design, or try a different drug completely.
Some researchers are hoping that the fruits of Moore’s law can help to curtail Eroom’s law. Artificial intelligence has already been used in drug discovery, assisting in the search for suitable disease targets and new molecule designs. Scientists are using Artificial Intelligence to manage clinical trials, including the tasks of writing protocols and recruiting patients.
