數位化｜由人到數據再回到人：激發創新的脈動 Innovation: From people to data to patients

被數據淹沒的世界，在2020年，全球總共產生了約50個皆位元組（ZB）──五百億兆位元組（TB）的數據。預計在2025年，這個數字將增長到175個ZB。隨著數據科學家運用演算法分析這個數據宇宙，並在各個人類努力的領域中獲得了全新的洞見，任何創新公司都不該忽視這個趨勢。

這當中包括了百靈佳殷格翰新成立的全球計算生物學和數位科學部門（gCBDS），正透過數據為藥物研發帶來更多益處。

「我們希望藉由這些發展開創的機會，補充研究組織現有的重要專業知識。」擔任gCBDS 全球團隊負責人的Jan Nygaard Jensen博士表示。Jensen博士與科學策略負責人Blaze Stancampiano共同制定百靈佳殷格翰領先該領域的組織路線圖。

數據、模式與新觀點

Stancampiano先生將其描述為對人類健康理解的新領域。他表示：「多虧對基因組的研究、生物庫和近幾十年的各項發展，我們能更深入了解人類的疾病。我們對基因和蛋白質的詳細知識，能使我們開發出效果更好、副作用更少的新藥物。」

Jensen博士指出，現在研究人可以已獲得龐大的醫學相關的人類數據，包括來自英國生物庫（UK Biobank）的50萬個參與者的詳細基因與健康資訊。

「與過往不同了，我們現在內部可以透過生物庫的合作獲得大量的人類數據，我們擁有足夠的計算能力，並為目前所需的基礎設施建立來分析這類數據。」Jensen博士說道。

數據科學家利用演算法搜索這些數據，搜尋模式一致的訊號，以獲取科學見解。Jensen 博士表示：「舉例來說，我們會透過研究數據中所有患有糖尿病的人是否擁有相似的基因組和蛋白質模式。」

基因是蛋白質生成的藍圖，而蛋白質則是構築身體的材料。蛋白質扮演著無數角色：作為抗體保護生物體、肌肉運動的肌球蛋白、提供皮膚和骨骼結構的膠原蛋白，以及作為激素、酵素等。透過適當影響基因和蛋白質，醫學科學能調控人體特定功能的運作，甚至修復身體的異常狀態。

數據與實驗室科學──相輔相成的組合

gCBDS團隊運用計算機輔助法尋找一系列基因和蛋白質（亦稱為「目標」），這些目標與某些疾病相互關連甚至具有因果關係。這些關聯性使人們更容易罹患特定疾病，並成為開發新藥療法的潛在候選目標。

然而，僅靠電腦分析無法確定這些假設是否指向正確的目標，在藥物開發的流程中，通過實驗室的多方驗證仍是必要的。「我們與創新單位的優秀生物學家、化學家和技術專家團隊合作，進一步評估特定模式和訊號是否參與了疾病的發展，我們的藥物組合做為新的治療目標是否合適。」Jensen 博士說道。

「合作絕對是成功的關鍵。」他說：「我們並非談論傳統實驗室工作與人工智慧間的矛盾，而是將它們結合起來，獲得過去無法發現的新見解。」

換句話說，跨領域的gCBDS團隊結合了人工智能和人類的洞察力。「我們訓練人工智慧來辨識模式，在大規模的數據中，它能做得比人類更好。」Jensen博士說：「但首先，我們必須利用我們對生物學的理解，來決定哪些模式是有意義的，這是人工智慧無法取代的部分。」

「我們透過計算和分析洞察力，將大規模的複雜疾病數據轉化為新型治療概念的發現，加速我們的藥物組合。」－Jensen博士，gCBDS 全球團隊負責人。

內部數據提供了優勢

照Stancampiano先生的說法，百靈佳殷格翰的目標是走在研究的最前端，提供75％的首創分子組合，其中50％對患者具有突破性的潛力，例如，美國食品藥物管理局對具有潛在顯著改善現有療法的藥物授予突破性療法的認證。

gCDBS部門有個明確的策略，是透過來自生物庫中以患者為中心的數據，與百靈佳殷格翰內部生成的廣泛實驗室數據相結合，來實現這一目標。「這使我們能夠更快地發現機會，進而成為第一個開發精確適應藥物的公司。」Stancampiano先生說道。

這正是gCBDS如何以符合百靈佳殷格翰使命的團隊工作：為新藥物和治療方法的開發奠定基礎，從而改善患者的健康和生活品質。

DIGITALIZATION

Innovation: From people to data to patients

Artificial intelligence, analytics and data science applications are rapidly influencing all parts of the medical research and development process. To incorporate these capabilities into its world-class research operations, Boehringer Ingelheim has created a new department in computational biology and digital sciences. The aim: accelerate the development of new types of medicines.

The world is awash in data. Around 50 zettabytes — 50 billion terabytes — were generated worldwide in 2020. That annual number is expected to grow to 175 zettabytes by 2025. As data scientists use algorithms to analyze this data universe to gain new insights in all areas of human endeavor, no innovative company can afford to be left behind.

That includes Boehringer Ingelheim, whose new Global Computational Biology and Digital Sciences department (gCBDS) is harnessing data for the benefit of drug discovery.

“We want to bring the opportunities from these developments to complement the significant existing expertise of the research organization,” explains Dr. Jan Nygaard Jensen, the Global Head of the gCDBS team. Together with Blaze Stancampiano, Head of Scientific Strategy, Dr. Jensen and his leadership team have developed a roadmap to establish Boehringer Ingelheim as a leading organization in this field.

Data, patterns and new perspectives

Mr. Stancampiano describes it as a new frontier in the understanding of human health. “Thanks to genome research, biobanks and other developments in recent decades, we now know much more about human disease,’’ he says. “We have much more detailed knowledge about genes and proteins which we can use to develop new medicines. This allows us to create more effective medicines with fewer side effects.”

Dr. Jensen notes the vast trove of medically relevant human data that is now available to researchers — including the UK Biobank, which contains in-depth genetic and health information from a half-million participants.

“Unlike a few years ago, we now have access to an enormous amount of human data, internally as well as externally via biobank collaborations“ Dr. Jensen says. “We have the computing power and are building the infrastructure we need to analyze this data. “

Data scientists use algorithms to search this data for patterns and consistent signals to derive scientific insight. “For instance,” Dr. Jensen says, “we investigate whether all people with diabetes in the database share a consistent genome and protein pattern.”

Genes are blueprints for the production of proteins, the building blocks of the body. Proteins perform countless roles: as antibodies to protect the organism; as myosins for muscle movement; as collagen to provide skin and bones with structure; as hormones, enzymes and much more. By influencing the right genes and proteins, medical science can influence specific functions — or malfunctions — in the human body.

Data and laboratory science — a synergistic combination

The gCBDS team uses computer-aided methods to find a series of genes and proteins (also referred to as “targets”) that have associations or even causal relationships with certain diseases. These associations could be responsible for putting people at increased risk of developing a specific disease — and making them potential candidates for a specific therapeutic approach to new medicines.

And yet, computer analysis alone is not enough to determine whether these hypotheses are pointing to the right targets. The entire array of proven processes for drug development and laboratory testing remains necessary. “We have an outstanding team of biologists, chemists and technology experts in the Innovation Unit who we partner with to further evaluate if specific patterns or signals are involved in the development of the disease, and ideally qualify as new drug targets in our portfolio,’’ Dr. Jensen says.

“Collaboration is absolutely key for success,” he says. “We’re not talking about a contradiction between traditional lab work and AI. We combine them to gain new insights which would never be discovered otherwise.”

In other words, the interdisciplinary gCBDS team weaves artificial intelligence and human acumen. “We train AI to recognize patterns, which it can do much better in largescale data sets than a human ever could,’’ Dr. Jensen says. “But first, we have to use our biological understanding to decide which patterns it even makes sense to look for. AI can’t do that for us, yet.”

“We translate large-scale complex disease data into NTC [novel therapeutic concept] discovery, by computational and analytic insight to accelerate our drug portfolio.”

Dr. Jan Nygaard Jensen, Head of gCDBS

In-house data provides an advantage

Boehringer Ingelheim’s goal is to be at the cutting edge of research and to deliver a portfolio of 75% first-in-class molecules, with 50% of them having breakthrough potential for patients, according to Mr. Stancampiano. The U.S. Food and Drug Administration, for example, grants breakthrough therapy designation to medicines that have the potential to be substantial improvements over available therapies.

By focusing on adding patient-centric data from biobanks to the extensive laboratory data generated within Boehringer Ingelheim, the gCDBS department has a clear strategy for achieving its goals. “This allows us to uncover opportunities more rapidly, which in turn enables us to be the first to develop precisely adapted medicines,” Mr. Stancmpiano says.

It's how the gCBDS team’s work fits into the broad Boehringer Ingelheim mission: to create the basis for new medicines and therapies that improve patients’ health and the quality of life.