Skip to Content
Artificial intelligence

Data-mining medieval text reveals medically bioactive ingredients

Medieval apothecaries used recipes with significant antibacterial properties, researchers say.

The Lylye of Medicynes is a 15th-century manuscript residing in the Bodleian Library in Oxford, England. It is a Middle English translation of an earlier Latin treatise on disease, containing case studies and treatment recipes. It was an influential text thought to have originally belonged to Robert Broke, personal apothecary to the English monarch Henry VI.

The Lylye of Medicynes is very familiar to historians who study medieval medical treatments. They have long known that some recipes contain ingredients, such as honey, with antibiotic properties.

But the broader question of the efficacy of medieval medicine in general is much harder to study. “The pharmacopeia used by physicians and lay people in medieval Europe has largely been dismissed as placebo or superstition,” say Erin Connelly of the University of Pennsylvania and colleagues from the University of Warwick in the UK.

Now that view looks set to change. Connelly and company say that medieval recipes followed a rational pattern of treatment that stands up to modern medical scrutiny. Their evidence comes from data mining the patterns of ingredients in the Lylye of Medicynes, which reveals networks of substances in the recipes with significant bioactive properties.

First, some background. The Lylye of Medicynes contains 360 recipes, each in a standard format that begins with the type of remedy (an ointment syrup or plaster, for example), then specifies the phase of illness when it should be applied, and ends with a list of ingredients.

Data mining this text is no easy task. The recipes mention over 3,000 ingredients for the treatment of 113 different conditions. Of these conditions, 30 describe symptoms such as broken skin, purulence, redness, black crust, foul smell, heat or burning, and so on, which translate into symptoms of external infections.

One challenge is that the text often refers to the same ingredients using different words and spellings. For example, the herb fennel is referred to as fenel, feniculi, feniculum, marathri, maratri, and maratrum. These must all be condensed under the same heading.

However, various parts of a plant can contain different active ingredients, and this also has to be taken into account. So root of fennel, juice of fennel, and seeds of fennel must all be included separately. The team also corrected spelling variants by hand.

Having standardized the ingredients, the team then studied the networks they formed. For this they created a node for each ingredient and drew connections between them if they appeared in the same recipe. The more often these ingredients appeared together, the stronger this connection became. Having assembled the network, the researchers used a standard algorithm to look for communities within the network.

The findings make for interesting reading. “The results clearly show the existence of a hierarchical structure within the recipes,” say the researchers.

Each community in the network is composed of smaller communities, all with a common kernel of ingredients. For example, one kernel of ingredients consists of aloe vera plus “sarcocolla nutria,” a gum from one of several Persian trees mixed with breast milk.

Several single ingredients play important roles in the network. These include honey, vinegar, and pomegranate blossoms.

The team’s next step was to search for emblematic recipes that exploit combinations of ingredients. They then searched the modern medical literature for evidence that those recipes could have worked.

For example, one recipe in the treatise is a mouthwash described as a treatment for “pustules, ulcers, apostemata (swelling/inflammation), cancer,  fistula, herpestiomenus (gangrene), and carbunculus (carbuncle; suppurating boil).”

This concoction is made with “sumac, galle, psidia (the rind of pomegranate or the bark of the tree), balaustia, mastic (resin exuded from the mastic tree, Pistacia lentiscus), olibanum, hony, and vinegre” probably mixed with nitrite or breast milk.

A significant question is whether any of those ingredients have antibacterial or immunomodulatory effects. To find out, Connelly and her colleagues looked them up in the Cochrane Database of Systematic Reviews, a well-known library of evidence-based medical research.

It turns out that there is good evidence that some of those ingredients are bioactive. Honey is known to have antibiotic properties, and the UK’s National Health Service regularly uses it for wound healing. Vinegar is a good disinfectant, and breast milk contains a variety of antimicrobial components. Bile (or galle) is also acknowledged as a potent bactericide.

However, there is scant evidence that aloe, frankincense, mastic, and sarcocolla have healing effects. For example, a Cochrane review of the wound-healing effects of aloe found that the relevant studies were generally of poor quality. So the jury is still out about the power of those substances.

Nevertheless, combining the specified ingredients in a single mouthwash clearly makes sense. It increases redundancy—if one ingredient doesn’t work, another might—and it “could increase efficacy against a particular target microbial species by attacking several cellular targets at the same time, or allowing for chemical activation of particular component molecules,” say Connelly and co.

They conclude that the mouthwash recipe, and others like it, reflect a rational approach to medical decision making.

That’s interesting work. It implies that the conventional view of medieval medicine as little more than hocus pocus needs to be rethought. “This work demonstrates the possibility to use algorithms from complex networks to explore a medieval medical data set for underlying patterns in ingredient combinations related to the treatment of infectious disease,” the researchers say.

Additionally, they believe there could be much more to discover in medieval texts, including the possibility of novel antimicrobial agents that are as yet unknown to modern science. “The use of digital technologies to turn these texts into databases amenable to quantitative data mining requires a careful interdisciplinary approach, but it could provide an entirely new perspective on medieval science and rationality,” say Connelly and co.

Ref: https://arxiv.org/abs/1807.07127  : Data Mining a Medieval Medical Text Reveals Patterns in Ingredient Choice That Reflect Biological Activity Against the Causative Agents of Specified Infections

Deep Dive

Artificial intelligence

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

OpenAI teases an amazing new generative video model called Sora

The firm is sharing Sora with a small group of safety testers but the rest of us will have to wait to learn more.

Google’s Gemini is now in everything. Here’s how you can try it out.

Gmail, Docs, and more will now come with Gemini baked in. But Europeans will have to wait before they can download the app.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at customer-service@technologyreview.com with a list of newsletters you’d like to receive.